Skip to main content

Full text of "Ethics. -"

See other formats



0^ -io9-S't^ 

Digitized by the Internet Archive 

in 2010 with funding from 

Lyrasis IVIembers and Sloan Foundation 




P. H. Nowell-Smith was born in 1914 and 
educated at Winchester and New College, 
Oxford. In 1937 he was elected to a Com- 
monwealth Fellowship in philosophy at 
Harvard University. During the war he 
served in the Army in the Middle East and 
in India. From 1946 to 1957 he was Fellow 
of Trinity College, Oxford, and he is now 
Professor of Philosophy at the University 
of Leicester. He has written and broadcast 
on a variety of topics. Professor Nowell- 
Smith is married and has four children. 




Penguin Books Ltd, Harmondsworth, Middlesex, England 

Penguin Books Inc., 3300 Clipper Mm Road, Baltimore 11, Md, U.S.A. 

Penguin Books Pty Ltd, Ringwood, Victoria, Australia 

First pubUshed 1954 
Reprinted 1956, 1959, 1961, 1964, 1965 

Copjrright © P. H. Nowell-Smith, 1954 

Made and printed in Great Britain 

by Cox & Wyman, Ltd 

London, Fakenham and Reading 

Set in Monotype Imprint 

This book is sold subject to the condition 

that it shall not, by way of trade, be lent, 

re-sold, hired out, or otherwise disposed 

of without the publisher's consent 

in any form of binding or cover 

other than that in which 

it is published 


Editorial Foreword 7 


1. The Task of Ethics ii 

2. Theoretical Ethics 23 

3. Intuitionism 36 

4. The Analogy between Ethics and Science 48 

5. The Logic of Adjectives 61 

6. The Logic of Sentences and Arguments 75 


7. The Purposes of Practical Discourse 95 

8. Reasons for Choosing 105 

9. Motives 122 

10. Egoism and Hedonism 133 

11. Advice and Exhortation 145 






'Right 'and 'Ought' 



Duty and Obligation 



Duty and Purpose 



The Purpose of Moral Rules 









Freedom and Responsibility (i) 



Freedom and Responsibility (2) 







Editorial Foreword 

MR P. H. nowell-smith's book on ethics is 
one of a series of philosophical works which are 
appearing in a similar form. The series consists 
mainly in original studies of the work of certain 
outstanding philosophers, but it covers also a 
number of more general topics, including, besides 
ethics, political philosophy, logic, the theory of 
knowledge, and the philosophy of science. 

The series is not designed to reflect the stand- 
point or to advance the views of any one philo- 
sophical school. Since it is addressed to an 
audience of non-specialists, as well as profes- 
sional philosophers, the contributors to it have 
been asked to write in as untechnical a manner 
as their subjects allow, but they have not been 
expected to achieve simplicity at the cost of 
accuracy or completeness. 

There is a distinction, which is not always 
sufficiently marked, between the activity of a 
moralist, who sets out to elaborate a moral code, 
or to encourage its observance, and that of a 
moral philosopher, whose concern is not pri- 
marily to make moral judgements but to analyse 
their nature. Mr No well- Smith writes as a moral 
philosopher. He shows how ethical statements 
are related to, and how they differ from, state- 
ments of other types, and what are the criteria 
which are appropriate to them. His book deals 
with the most important problems in the field 
of moral philosophy, from the objectivity of 
values to the freedom of the will. 


Theory and Practice 


The Task of Ethics 


A BROAD distinction may be drawn between theoretical and 
practical sciences. The purpose of the former is to enable us 
to understand the nature of things, whether the things be stars, 
chemical substances, earthquakes, revolutions, or human be- 
haviour. These sciences consist in answers to such questions 
as ' What is an acid ? ', ' What are the laws of planetary motion ? ', 
'How do bees find their way about?', 'Why does wood 
float and iron sink?', 'What are the marriage laws of the 
Arapesh?'. The answers take the form of statements, descrip- 
tions, generalizations, explanations, and laws. I shall call such 
discourse ' theoretical ', ' fact-stating ', or ' descriptive ' discourse ; 
but it must not be supposed that every sentence in such dis- 
course is a theory or states a fact or describes something. 
Newton's laws belong to descriptive discourse, but they do 
not describe anything. 

Practical discourse, on the other hand, consists of answers 
to practical questions, of which the most important are 'What 
shall I do ?' and 'What ought I to do ? '. If I put these questions 
to myself the answers are decisions, resolutions, expressions 
of intention, or moral principles. If I put them to someone else 
his answer will be an order, injunction, or piece of advice, a 
sentence in the form ' Do such and such'. The central activities 
for which moral language is used are choosing and advising 
others to choose. 

Traditionally, moral philosophy has always been regarded 
as a practical science, a 'science' because it was a systematic 
inquiry the goal of which was knowledge, and 'practical' 
because the goal was practical knowledge, knowledge of what 
to do rather than knowledge of what is the case. 



The words 'morals' and 'ethics' are derived from words 
meaning 'customs' or 'behaviour'; but the role of the moral 
philosopher was never conceived to be that of describing or 
explaining the customs or behaviour of men. That is the task 
of the psychologist, sociologist, anthropologist, historian, 
dramatist, and novelist. Moral philosophers set out to perform 
different tasks. The first was to answer practical questions, the 
second to criticize, evaluate, or appraise customs and beha- 
viour. They claimed, not to tell you what men do, but to tell 
you which of the things that men do are good and which are 
bad. This second type of judgement, which is expressed in 
verdicts or appraisals, seems closer to theoretical than to prac- 
tical discourse. To say that Jones is a good man is not - on the 
face of it - to tell anyone what to do, but rather to tell you what 
sort of man Jones is. It states a fact and from its grammatical 
form alone v/e should conclude that it is more like 'Jones is a 
tall man' than like 'Do what Jones does'. 

But the great philosophers of the past always treated ques- 
tions of appraisal as subordinate to practical questions. They 
assumed - and who would not ? - that the point of telling you 
that Jones is a good or a bad man is that you should imitate or 
not imitate Jones, that you should or should not give Jones the 
job or do whatever else might be in question. When they de- 
picted the Good Life they would have thought it very odd if 
someone had said: "You have told me what the Good Life is 
and I agree with everything you say. Now tell me what I 
ought to do". Such a man has obviously misunderstood the 
philosopher's talk about the Good Life. For this talk was never 
intended to be a description of anything ; it was from the start 
assmned to be an injunction to do something, to adopt this or 
that course, to subscribe to this or that moral code. 

Moral philosophers did not, of course, undertake to give de- 
tailed practical advice as to how you should behave on this or 
that occasion. A philosopher is not a parish priest or Universal 
Aunt or Citizens' Advice Bureau. As we shall shortly see, 
different philosophers held very different views as to the way 
in which moral philosophy can help you to answer practical 



questions. But they all agreed that the goal of moral philo- 
sophy is practical knowledge, not that we should know what 
goodness is but that we should become good. From Plato, 
Aristotle, and Epicurus, from Hobbes, Spinoza, and Butler you 
can learn how, in their opinion, you ought to live. 

And these philosophers were agreed also on another impor- 
tant point. They take for granted the fact that men have cer- 
tain aims, purposes, and desires which they wish to achieve, 
fulfil, and satisfy. The achievement of these aims is variously 
called 'The Good Life', 'The Good for Man', 'Happiness', 
and ' Felicity' : and the task of the moral philosopher was to de- 
pict this state in broad outline and to tell you how you can 
achieve it. Although Plato starts his most elaborate treatise on 
moral philosophy with the question 'What is Justice?', the 
principal question with which he is concerned is ' Which of the 
two (justice or injustice) will bring happiness to its possessor ? '.^ 
The notion of duty does not play the central role in traditional 
that it plays in modern ethics and the notion of doing one's 
duty for duty's sake hardly appears before Kant. Earlier philo- 
sophers thought it quite sensible.,to ask 'Why should I do my 
duty ? ' ; the obligation to do one's duty needs justifying and 
can only be justified by showing that doing his duty is, in the 
short or long run, advantageous to the agent ; indeed the classic 
treatises on the subject might be said to be mainly concerned 
with this justification. This point of view is called 'teleolo- 
gical' and is opposed to that called 'deontological', according 
to which duty rather than purpose is the fundamental concept 
of ethics.2 

If we turn to the great religious systems of the world we 
find the same emphasis on the practical nature of moral ques- 
tions, the same assumption that life has a goal, be it the Chris- 
tian or the Moslem Paradise or the Buddhist Nirvana, and the 
same assumption that the rules we are enjoined to live by are 
rules for achieving this state. This assumption is of the 

1. Republic, a^zi d. 

2. From the Greek ' telos ' meaning ' purpose ' and ' deon ' meaning - 
roughly - 'duty'. 



greatest antiquity. The Egyptians had a word 'Ma'at' which is 
translated in three different ways. It means (a) 'being straight, 
level, or even', (b) 'order, conformity, regularity', and (c) 
'truth, justice, righteousness'. It has obvious affinities to our 
word 'right'. Now the earliest moral document we possess is 
a manual of instruction on good behaviour to budding civil 
servants. In this the aspiring official is enjoined to follow the 
rules of Ma'at, but only because, by so doing, he will get on 
in the world. In later times the fruits of Ma'at were thought 
to lie in the next world rather than in this, but the basic con- 
ception is the same. We are to practise virtue because, in the 
short or long run, it pays.^ 

In Christian ethics we find the same basic assumptions. 
"What shall I do", asks the lawyer, "to inherit eternal life?"^ 
The difference between this question and Plato's is simply that 
felicity is now thought to lie in the next world, not in this. No 
question is raised as to whether the lawyer wants to inherit 
eternal life; this is treated as a datum of the problem. Still less 
is there any question as to whether he ought to want this or 
ought not rather to have some other aim. In Christian, as in 
Greek ethics, there is no suggestion that moral rules ought to 
be obeyed irrespective of any purpose which will be served by 
obeying them. The answer does indeed take the form: "What 
is written in the law? How readest thou ? ". But this would not 
be an answer at all if it were not assumed that the rules con- 
tained in the law are to be obeyed because obedience is con- 
ducive to what the lawyer wants to achieve. Without this 
assumption the answer is simply irrelevant to the lawyer's 

But to say that moral rules are rules for achieving happiness 
or the Good Life does not tell us much. In what does happiness 
consist ? How can we tell a good life from a bad one ? How do 

1. H. Frankfort and others: Before Philosophy, Ch. IV. 

2. Luke X. 25. 

3. It might be thought that to look at Christian ethics in this way is 
to give it an unplausible and shocking egoistic twist. On this see below, 
chapter 10. 



we know whether obedience to the proffered rules is really 
going to lead to happiness ? It was to these very general ques- 
tions that the great philosophers mainly devoted themselves. 
In order to answer them they had to range over a wide field. 
Since man is a social animal, the Good Life must be life in 
society, and politics must be discussed ; indeed ethics and poli- 
tics were, for these philosophers, one subject. Since it is idle to 
tell men what they ought to do unless we know what men are, 
psychology also comes into the picture, and philosophers spent 
much time analysing the human soul. It is also, for reasons to 
be given later, necessary to study logic; though this was more 
apparent to some philosophers than to others. 


Students of ethics are apt to be disappointed to find that, 
although the subject has been studied for over two thousand 
years, it does not seem to have produced any established sys- 
tem of truths comparable to those of mathematics and the 
natural sciences. Why is Aristotle's Ethics still worth reading, 
while his Physics is of interest only to scholars and historians ? 
If moral rules are rules for achieving happiness, it ought to be 
fairly easy to discover which rules are good and which are bad, 
which to adopt and which to reject. 

Part of the answer to this criticism of moral philosophy lies 
in the fact that as soon as any progress is made in solving any 
general type of problem that problem ceases to be a philoso- 
phical one and becomes a problem for someone who is called, 
not a 'philosopher', but a 'scientist'. Suppose, for example, 
that a man wants to bring up his children as well as possible. 
He is faced by two quite different types of problem, which we 
may call 'problems about ends' and 'problems about means'. 
(fl) What sort of person do I want my child to become ? and 
(&) What method of up-bringing is most likely to turn him into 
this sort of person ? 

In answer to the first question it is no use saying that he 



wants his child to lead a good and happy life. Of course he does. 
But in what do goodness and happiness consist ? What, in more 
detail, is the ideal life for him - or for anyone ? These are philo- 
sophical questions because the ability to answer them is not a 
matter of special expertise. But if we turn to the second type of 
question the position is different. 

Suppose that he has decided that he does not want his child 
to be grasping and aggressive. The question 'What sort of up- 
bringing prevents people from becoming grasping and aggres- 
sive ? ' is a question of fact, the ability to answer which could 
be a matter of expertise. If it is true that early weaning makes 
people grasping and aggressive, which is something that psy- 
chologists might be able to establish empirically, that settles 
the matter. Psychology is a young science ; but enough is now 
known to make many questions about the best method of 
achieving an end 'scientific' questions. That they continue 
to be treated as philosophical or moral questions is a matter 
for regret. 

But is the man's problem really solved when he has learnt 
the facts about weaning? It might not be. Suppose that his 
wife is so ill that it would be dangerous for her to go on suck- 
ling the child. He is now faced with a problem of the first kind 
again. For he now has to decide, not between two methods of 
achieving a given end, but between two different ends consist- 
ing in the effects on both his wife and his child of either 
weaning or not weaning. And no ' science ' can help him solve 
this problem ; he has to discover which of these two complex 
states he really wants. 

The second part of the answer is of greater philosophical 
interest. Why cannot we discover a body of general moral 
truths that would help us to solve particular moral problems in 
the way that geometrical truths help the surveyor and mech- 
anical truths help the engineer to solve their practical prob- 
lems ? Some philosophers have thought it possible to discover 
such truths., Plato, for example, seemed to think that, although 
it would be very difficult and no one had yet done it, it might 
be possible for a man to come to know The Form of the 




Good; and this would give him an insight into moral rules not 
unlike the insight of the geometer into geometrical axioms. 
Like Plato, Hobbes was much impressed by the method of geo- 
metry and he thought that moral rules were Rules found out 
by Reason for avoiding social calamity; and Spinoza frankly 
casts his ethics into a deductive, geometrical form. 

But the analogy with mathematics or with any of the 
natural sciences will not do. I shall try later to explain in 
detail why it will not do; here it must suffice to notice that 
the conditions which make the deductive method so fruitful 
in the sciences do not all obtain in the field of conduct. The 
success of the sciences is in part due to the possibility of 
discovering functional relations between measurable quan- 
tities and in part due to the possibility of giving precise 
meanings to the words employed. The classical Utilitarian 
theory was an attempt to produce a sort of mechanics of 
behaviour; and it failed just because these necessary condi- 
tions do not obtain. Even if we neglect the fact that ethics 
raises questions about ends and take the duty to produce the 
maximum pleasure all round us as a datum, it is impossible 
to measure amounts of pleasure in the way that we can measure 
amounts of heat or energy; and the duty to produce the 
maximum of pleasure all round can certainly not be accepted 
as a datum unless 'pleasure' is construed so vaguely as to 
be useless. 

Moreover the view that no progress has been made in 
solving the fundamental problems of ethics is an illusion 
based on the fact that some of these problems were solved 
very early in the history of mankind. Both the practical diffi- 
culties of deciding what to do and the difficulties encountered 
in theoretical ethics are unlike those encountered in mathe- 
matics and science. In practice difficulties arise mainly from 
the conflict of moral rules and in the application of well- 
known rules to particular cases ; in theoretical ethics they 
are, as we shall see, mainly difficulties of understanding the 
logical behaviour of and relations between the concepts used 
in practical discourse. 



Problems of conduct arise on particular occasions, and 
difficult problems arise only when the circumstances are un- 
familiar. Indeed they can only so arise, since we all know what 
to do when the case is covered adequately by some familiar 
rule. Sceptical doubts about the validity of our moral code 
or its application to familiar circumstances reflect no genuine 
uncertainty about what we ought to do. 

There are a few fundamental rules of conduct that have never 
changed and probably never will ; indeed it is difficult to ima- 
gine what life in society would be like if we abandoned them. 
The more we study moral codes the more we find that they do 
not differ on major points of principle and that the divergencies 
that exist are due partly to different opinions about empirical 
facts, for example about the effects of certain types of conduct, 
and partly to differences in social and economic organization 
that make it appropriate to apply the fundamental rules now 
in one way, now in another. Thus all codes agree that we have 
a duty to requite good with good; but obedience to this rule 
will involve behaving in ways that will differ according to the 
view that a society takes of what it is to do good to someone. 
In some societies it is rude for a guest to eat everything on his 
plate, in others it is rude for him not to do so ; but both agree 
in enjoining that we should not be rude to our hosts. 

It is not enough to know the general, unchanging rules ; we 
must also know how to apply them. And it is for this reason 
that moral rules, like the law, cannot be codified for all time. 
One of the main tasks of the lawyer is to apply well-tried and 
stable general principles to cases that could not have arisen in 
an earlier age simply because the facts involved could not have 
arisen. New social and economic relations between man and 
man give rise to new rights and duties that could not have 
been contemplated by the authors of a particular moral or 
legal code. 

The idea of a 'scientific morality' - if it means anything 
more than the laudable recommendation to make use of dis- 
coveries in psychology when we are thinking about means - is 
as chimerical as the ideal which inspired the Code Napoleoti. A 



detailed moral code, a sort of handbook to which we might 
turn for the answer to every moral problem, cannot help us, 
because difficulties will always arise about the application of 
the rules to new cases and because the cases in which the need 
for practical thinking is particularly acute are just those which 
are new and those in which we suspect that there is some good 
reason for breaking the accepted code. The need to think afresh 
about moral problems is ever present and particularly great in 
a period of rapid economic and social change and rapid advance 
in knowledge of human nature. Most of our detailed rules were 
evolved in societies very different from our own and by people 
who knew far less about human nature than we now do. This 
is particularly true in matters of sexual morality, where the 
rules are, or were, considered to be hypothetical and justified 
by reference to the consequences of obeying them to the indi- 
vidual and to society. If we know, as we now do, that obedi- 
ence to some of these rules does not promote the desired end, 
we must either abandon them or make them out to be absolute, 
categorical rules of morality, which in many cases is very 


The impossibility of a 'scientific morality' and the reasons why 
it is impossible were more clearly understood by Aristotle 
than by any other philosopher. Like other philosophers he be- 
lieved that the aim of moral philosophy was to make men 
good, but he did not think that he could make them good by 
lecturing on ethics or by writing a handbook. He thought that 
the help that moral philosophy can give was of a more indirect 
and subtle kind ; and since the scope and method of this book 
are of a more Aristotelian than Platonic kind, it is necessary to 
explain what this is. 

Suppose someone were to ask me to give him a moral code 
to live by, I should reply - as Aristotle in effect did - "I can't 
give you a code ; go and watch the best and wisest men you can 



find and imitate them". I should not have given him what he 
asked for ; but I should have given him the sort of help that a 
moral philosopher can give. If I went on to give the reasons 
for this advice, I should help him by explaining that he has 
put his question in the wrong way. What he really wanted was 
practical knowledge; he wanted to know how to live. But he 
had jumped to the conclusion that this sort of knowledge could 
be contained in a moral code. It is as if someone who wanted 
to learn how to play the piano were to ask for a manual of in- 
structions and to suppose that when he had mastered the 
manual he would be able to play. Now whether morality can 
be learned from books or only by practice is a typical philo- 
sopher's question. I am not claiming to give the answer - which 
is in this case obvious - but only to use this as an illustration 
of the sort of help that a philosopher can give. 

Some of our practical difficulties arise because we are not 
sure about the facts of the case. But we have other difficulties 
which are of a kind that is not so easily detected. These are j 
due to our inexpertness in handling the concepts or words that ' 
\ye use for solving practical problems. We have a specific 
vocabulary for dealing with moral questions. It contains such 
words as desire, appetite, will, voluntary, choice, approval, 
conscience, remorse, guilt, deserve, pleasure, pain, duty, obli- 
gation, good, and evil. These, unlike the technical terms of a 
science, are not the property of specialists ; in a way, everyone 
understands what they mean; and people who do not knov/ 
what they mean, who do not know, for example, what pleasure 
or remorse is, will never learn anything from ethics. 

But there is another way in which people who are unversed 
in philosophy do not understand these words. They do not 
always know the coimexions between these words or between 
the ideas which they express. Under what precise conditions 
do we hold a man responsible ? Why do we punish thieves and 
not kleptomaniacs? What is the connexion between respon- 
sibility and deserving punishment? How is choosing related 
to wanting ? How does duty come into the picture ? Is it neces- 
sary to justify obedience to a rule of duty ? Does it even make 



sense to try to do so ? All these are typical questions of moral 

And there is another range of questions that are even more 
obviously questions about the meanings of words. How are 
these different concepts related to each other? Can 'good', for 
example, be defined in terms of ' pleasure ' or ' desire ' ? Can it 
be defined at all or is it an irreducible moral term ? 

There is a similar range of questions also about sentences. 
Do all the sentences that appear to do so state facts ? If we use 
some sentences to state facts, others to express decisions, 
others to give advice, issue orders and so on, how are all these 
things that we do with bits of language related to each other ? 
Earlier I drew a broad distinction between questions about 
ends and questions about means. Is this distinction as clear-cut 
as it seems to be ? And is this terminology of means and ends 
adequate to the subject, or are there some questions which 
stubbornly insist on posing themselves but which cannot be 
satisfactorily posed in this form ? 

This set of questions raises another set of an even higher 
order of abstraction, still more remote it would seem from 
solving a moral problem. If we distinguish between different 
types of question or sentence we presuppose that we have 
some basis for the distinction. What is this basis ? Do we dis- 
cover the difference by inspection ? And, if so, of what ? Their 
grammatical form ? Or by reflecting on the different jobs that 
they are used to do ? 

But, while the first range of questions that I have m( 
tioned has always occupied the attention of philosopher^ 
more obviously logical questions about words and s*^ 
were, until very recently, largely neglected. This w 
it was implicitly held that the logic of every type o 
must be identical. Grammarians might be interested . 
forms, statements, questions, commands, wishes, a*" 
But the province of philosophy was Truth and th 
of Truth was thought to be the 'proposition', ex 
indicative sentence which ascribes a 'quality' to 
Other moods and sentence-forms and other uset- 



sentences might be put, however important they might be h 
other ways, were irrelevant to the quest for Truth. In th 
following chapters I shall try to show how adherence to thi 
logical dogma has prejudiced and distorted the accounts whicl 
moral philosophers have given of what it is to make a mora 
decision or judgement. And if this examination involves post 
poning my own account of the logic of moral words, my excust 
is that a philosopher can only make his own views clear b] 
contrasting them with those of his predecessors. 



Theoretical Ethics 


In the last chapter I suggested that, to the great moral philo- 
sophers of the past, ethics was primarily concerned with an- 
swering practical questions. We sometimes find these questions 
put in what looks like a theoretical form : ' What is the Good for 
Man?', 'What is the nature of Goodness or of Obligation?'. 
Their investigations then take on a superficial resemblance to 
theoretical inquiries, inquiries into the nature of molluscs or 
carbohydrates or igneous rocks or conic sections. But the dis- 
guise is thin. When Plato asks "What is Justice?", it is clear 
that he keeps his eye continually on the question 'What ought 
we to do?'. 

Moral philosophers were, of course, always partly theor- 
eticians. They claimed to give a true account of what it is to 
make a moral judgement, to decide, deliberate, and choose, as 
well as to answer moral questions in a more direct way. But 
they did not, except superficially, represent moral judgements 
as being themselves theoretical statements, as being descrip- 
tions or explanations of a special world of moral qualities and 
objects. When, for example, they offered definitions of good- 
ness or obligation, they would claim that their definitions 
truly and faithfully reflected our ordinary use of the words 
'good' and 'ought', but they regarded judgements in the form 
'X is good' and 'X is obligatory' as practical judgements. 

When we turn, however, to the works of some of the best- 
known twentieth-century moralists we find this conception of 
moral judgements as practical judgements deliberately aban- 
doned. The direct object of ethics, we are told, is not Practice 
but Knowledge. And we shall soon see that this knowledge 
is not theoretical knowledge about moral judgements and 



concepts. Moral -judgements are themselves bits of theoretical 
knowledge about special moral objects. Knowing that this is 
right or that that is wrong is knowing that something is the 
case, not knowing what to do. 

This contrast between Practice and Knowledge implies the 
assumption that there is no such thing as Practical Knowledge, 
an assumption that the older philosophers would have re- 
jected if it had ever occurred to them. The moral philosopher's 
task is now conceived, not to be one of conducting a theoretical 
inquiry into practical wisdom, but to be one of investigating 
questions, judgements, doubts, and beliefs that are themselves 
theoretical. The moral philosopher not only makes theoretical 
statements about his subject-matter; his subject-matter con- 
sists of theoretical statements. 

And it is partly in consequence of this change of view as to 
what sort of a subject moral philosophy is that the intuitionist 
philosophers whose theories I am about to examine have 
accused traditional moral philosophy of resting on a mistake. 
This is a sweeping condemnation and we must now see how it 
came to be made and whether it is just. 

According to Professor Broad "Ethics may be described as 
the theoretical treatment of moral phenomena ".^ This ap- 
proach at once assimilates the moral philosopher's task to 
that of a scientist. The underlying assumption is that there is 
a special field or subject-matter, a special set of objects, qual- 
ities, or phenomena that the moral philosopher is going to study 
in the way that the chemist studies the chemical properties of 
matter, the geologist rocks, and the astronomer stars. 

It is true that an important distinction is soon drawn be- 
tween moral philosophy and natural science, and intuitionists 
have, as we shall see, a special way of marking this distinction. 
But it is important to insist on the implied analogy; for it is 
the uncritical use of this analogy that distorts what intuitionist 
writers have to say about the difference between the two 
subjects and leads them to their sweeping condemnations of 

I. Broad: Philosophy, 1946. Reprinted in Feigl and Sellars: Readings 
in Philosophical Analysis, p. 547. 



traditional ethics. In the next three chapters I shall argue that 
the analogy cannot be sustained and that intuitionists funda- 
mentally misrepresent the differences between moral and em- 
pirical discourse, differences which they recognize to be 
genuine and important, by treating moral discourse as if it 
was itself descriptive and explanatory. They have treated 
moral discourse as being descriptive of a special world of 
special objects, qualities, and phenomena, when the real burden 
of their arguments is to show that moral discourse is not 
descriptive at all. 

What, we must ask, are the objects that ethics studies, the 
objects that bear the same relation to the moralist as birds 
bear to the ornithologist or coins to the numismatist? We 
might in the past have pointed to pleasure and pain, to the 
various desires, aversions, and motives, that play a part in 
human conduct. But this will not do. The description, classi- 
fication, and explanation of these 'phenomena' belong to the 
empirical science of psychology, and psychology is not ethics. 
It is a cardinal doctrine of the intuitionists that ethical con- 
cepts cannot be identified with or reduced to psychological 
ones or to the concepts of any other natural science. 

According to Broad the special phenomena to be studied 
are of three kinds, moral judgement, moral emotion, and moral 
volition. Of these he gives the following examples : 

(a) I know or believe that I ought to keep a promise . . . 
and that it is wrong to inflict useless pain. These bits of moral 
knowledge or belief are instances of Moral Judgements. 

(b) I feel remorse or self-disapproval, as distinct from mere 
fear of punishment or embarrassment at being found out. 
These feelings will be instances of Moral Emotion. 

(c) In so far as I am influenced in my decision (between 
two alternative courses of action) by the thought that one of 
them is right, this is an instance of Moral Volition.^ 

We may concentrate on moral judgement since moral emotion 
I. Loc. cit. 



and volition owe their place in ethics to their connexion with 
moral judgement. Remorse seems to me to differ, as an 
emotion, in no fundamental way from embarrassment; it be- 
longs to ethics solely because this emotion is not called 
'remorse' unless it arises from the judgement that what one 
did was morally wrong. Similarly, volition is expressly said 
only to be ' moral ' if my decision is influenced by the thought 
(i.e. judgement) that the action is right. 

The central phenomenon of ethics is, then, the moral judge- 
ment or, as Broad also calls it, the "opinion that something 
is right or wrong". It should be noticed from the start thatj; 
moral judgements are said to be, not emotional reactions or I 
attitudes or expessions of approval and disapproval, but 
opinions. This word immediately takes us into the vocabulary' 
in which we talk about questions of fact, in science and inl 
everyday life. Opinions are true or false, correct or mistaken; 
a man can be convicted of error in holding an opinion in a way 
that he cannot if he merely expresses his taste. It may be 
'wrong' in some sense to like comic strips or the music of| 
Cole Porter; but we could not say that a man who liked; 
them was mistaken ; and if we did we should have to add that 
we were using ' mistaken ' in a different way from that in which 
we use it when we say that a man who thinks that New York 
is the capital of the United States is mistaken. 

There is nothing logically surprising in the notion of the 
whole world being mistaken about a point of fact ; indeed we 
know that for centuries everyone believed the earth to be flat ; 
but the suggestion that if everybody called strawberries ' nice ' 
they might all be mistaken is very odd indeed. The contrast 
between cases where a man can be said to be mistaken and 
cases where he cannot is usually put in the form of saying that 
matters of fact are 'objective', matters of taste 'subjective'. 
By using the word 'opinion' Broad shows that moral judge- 
ments are, for him, judgements of fact, not judgements of 
taste, objective not subjective. 

But if we say that the phenomena studied by the moralist 
are moral judgements or the reports of the moral conscious- 




ness, we are met by a peculiar difficulty. How does the moral 
philosopher 'theoretically treat' these? It would seem that, 
if the analogy with the natural sciences is to hold, he must ask 
such questions as : ' Under what circumstances does such and 
such a moral judgement occur?', 'What causes does it have?', 
'What effects?'. His job must be to classify moral judgements 
into types and genera and to discover general laws governing 
the occurrence of these 'phenomena'. But it is obvious that 
such an inquiry would be a part of empirical psychology and 
that the moralist is not interested in describing, classifying, 
and explaining these judgements. He is interested in dis- 
covering whether or not they are true. Judgements and 
opinions (usually dignified with the name of 'theories' when 
they are of a general kind) play an important part in science; 
and the student of a science must study them in the sense that 
it would ill beseem him not to be au fait with the theories 
current in his subject. But, in a quite different sense, the 
student of chemistry studies, not the theories of Dalton and 
Mendeleev, but the properties of chemical substances; these, 
and not the theories, are the objects or phenomena that he 
theoretically treats. And we soon find Broad taking it for 
granted that the phenomena studied by the moralist are not 
moral judgeme?its at all, nor even moral emotions and volitions, 
but a special set of objects, namely the Right, the Wrong, 
the Good, and the Bad. 

If we took Broad's opening paragraph literally we should 
have to regard the moral philosopher as a historian of moral 
opinions analogous to the historian of natural science whose 
objects of study are indeed the theories held at different times 
by scientists. Professor Westermark does in fact take this 
line; he believes that the purpose of ethics "can only be to 
study the moral consciousness as a fact".^ But he does this- 
frankly as an historian of morals because he believes that there 
is nothing to study but the psychological and sociological 
phenomena involved. 

Sir David Ross also takes as his starting point "the existence 
I. Ethical Relativity, p. 6i. 


of what is commonly called the moral consciousness, . . 
the existence of a large body of beliefs and convictions to the 
effect that there are certain kinds of acts that ought to be done 
and certain kinds of things that ought to be brought into 
existence".^ But it is clear that, for him, we 'start from' 
these beliefs not in the sense that they are the phenomena 
that we study first, but in the sense that the student of any 
subject 'starts from' the opinions of his teachers. There is, 
however, a difference between ethics and a natural science 
which Ross notices without, I think, realizing its crucial im- 

In the natural sciences, he says "it would be a great mistake 
to take as our starting point either the opinions of the many 
or those of the wise. For in them we have a more direct 
avenue to truth; the appeal must always be from opinions to 
the facts of sense-perception ".^ Clearly Ross does not mean 
that it would be a mistake for a student to start from the 
opinions of his teacher. How else could he start ? Nor does he 
mean that it would always be a mistake for a layman to prefer 
the verdict of an expert to his own untutored observation. He 
means that in the natural sciences and, indeed, wherever the 
question is one of fact there is a test of truth superior to the 
appeal to authority. This is a consequence of the logic of the 
word 'fact'. It would be foolish for the young scientist to 
ignore the opinion of the expert and foolish for a traveller to 
ignore the information contained in Bradshaw. But Bradshaw's 
statement that the train leaves at ten o'clock is not true be- 
cause Bradshaw says it; it is true because the train leaves at 
ten o'clock. The opinions of eminent scientists are ignored at 
our peril; but if they are true it is because they correspond 
with the facts, not because eminent scientists hold them. 

Ross goes on to say that in ethics we have no direct avenue 
to truth such as that which observation provides in empirical 
tnatters, and must therefore be content with the opinions of 
the many and the wise. But our inability to observe ethical 
truth has far-reaching consequences. It must raise a doubt 

I. The Foundations of Ethics, p. i. 2. Op. cit., p. 3. 



about the propriety of taking over into ethics the concept of 
truth and its attendant concepts such as 'fact', 'false', 'mis- 
taken', 'correct', and so on. For, in empirical matters, our use 
of these concepts is, as Ross says, bound up with our having 
the direct avenue of observation. 

Now it is a fact that we do talk of truth and falsity in moral 
matters, of opinion and knowledge (not just emotion and 
taste), of being correct and mistaken; and this is a fact that 
must be accounted for. But the logic of this talk is now seen 
to be different in an important way from its logic in empirical 
discourse. The two uses are analogous ; but the analogy is not 
exact. And this should make us cautious about applying to 
ethics logical concepts that we customarily apply to empirical 
discourse. When, in an empirical case, I have not observed 
something myself but accept someone else's opinion as true, 
I do so because I believe that it corresponds with the facts and 
that he has observed this although I have not. But, if Ross is 
right, in ethical matters the many and the wise are, like me, 
debarred from the direct access to truth that we have in em- 
pirical matters. 

And this leads us to ask what it is in ethics that makes one 
man's opinion true and another's false, when there is nothing 
analogous to the scientist's appeal from opinions to the facts 
of sense-perception. The question is a crucial one, but what 
Ross has to say on it seems to me obscure. "We must start 
with the opinions that are cyrstallized in ordinary language 
and ordinary ways of thinking, and our attempt must be to 
make these thoughts, little by little, more definite and dis- 
tinct, and by comparing one with another to discover at what 
points each opinion must be purged of excess and mis- 
statement till it becomes harmonious with other opinions that 
have been purified in the same way. "^ Part of this programme 
is not difficult to understand. We can see how careful thinking 
will make our thought definite, distinct, and harmonious. But 
how are we to detect a mw-statement ? How are we to tell 
when the rejection of an opinion is really the purging of a 
I. Loc. cit. 



falsehood and not the rejection of a truth ? Ross seems to be 
advocating a Coherence theory of ethics in which the truth of a 
moral opinion would depend on its coherence with other 
opinions. But this is not really so. For we soon find that an 
appeal analogous to the scientist's appeal to sense-perception 
is introduced and the test of truth remains one of correspond- 
ence. A direct avenue is discovered and it turns out to be a 
very short one. We are directly aware of the moral pheno- 
mena to which true moral opinions correspond. 

The method actually practised by intuitionist philosophers 
is partly that described by Ross in this passage. They take the 
tangled web of our ordinary moral discourse, which they 
believe to be often obscure and muddled, and subject it to 
careful analysis. Vague and general statements expressed in 
loose terminology are split up into a number of different and 
precise things that they could possibly mean and these products 
of analysis are submitted for our inspection. But how do 
we know which of them is true ? We may compare them with 
similar products presented by rival moralists; but which are 
we to accept ? 

In the course of the intuitionist's argument a crucial, but 
unnoticed, shift of interest has occurred. The ' phenomena ' to 
be studied were originally said to be men's moral volitions, 
emotions, and opinions, of which opinions were the most im- 
portant. But they are now seen to be, not the opinions, but 
what these opinions are about, the special objects or proper- 
ties denoted by the words in our moral vocabulary, the Right 
and the Good. The belief that ethics is a theoretical science, 
that its task is the description, classification, and explanation 
of special phenomena, objects, and qualities, is maintained in 
spite of the fact that the intuitionist has noticed a crucial 
difference between ethics and the other theoretical sciences. 
The difference was that, while in the sciences opinions are 
checked by reference to facts, in ethics the opinions themselves 
are the only facts available. Yet this conversion of ethics into 
a sort of empirical psychology is so unplausible that the in- 
tuitionist abandons it from the start and, to save the analogy 



with science, draws our attention to (or postulates) a special 
set of objects which are to the moralist what rocks, stars, and 
birds are to their respective scientists. The appeal is made to 
observation ; but it is observation of a very special kind. 


In rejecting the evolutionary theory of ethics, (according to 
which 'right' means 'more evolved') Ross says: "There is 
really no resemblance between the characteristic which we have 
in mind when we say 'right' or 'obligatory' and that which 
we have in mind when we say 'more evolved'. ... If we ask 
ourselves what 'more evolved' means, we shall find in it, I 
think, two main elements: (i) that conduct so described 
comes, in time, after a process of evolution of more or less 
duration, and (2) that it has a characteristic which usually 
emerges in the course of evolution, that of being complex, in 
comparison with the simple activities which appear in an early 
stage of evolution. And it is surely clear that neither temporal 
posterity nor complexity, nor the union of the two, is that 
which we mean to refer to when we use the term 'right' or 

Ross's criticism of the attempt to define 'right' in terms of 
appearing at a later stage of evolution is surely correct ; but I 
have italicized certain phrases in order to draw attention to 
the fact that Ross's way of putting his point involves certain 
assumptions about the logical behaviour of moral concepts. 

(a) The words 'right' and 'obligatory' refer to certain 
characteristics. (Other philosophers use 'denote' and 'stand 
for' as synonyms for 'refer to', and 'property' and 'quality' as 
synonyms for 'characteristic'.) 

(b) Clarifying the meaning of a word is represented as 
examining our moral consciousness and discovering therein 
the characteristic to which the word refers. We have this 

I. Op. cit., p. 13 (my italics). 


characteristic 'in mind' and, if we are careful, we can see it 

(c) Discovering whether or not two words mean the same 
thing is a matter of examining the characteristics for which 
they stand and just seeing whether or not these characteristics 
are identical. 

(d) In other passages Ross represents apprehending a 
moral truth as 'seeing' whether or not a certain object has a 
certain characteristic. 

Putting these points together we find that in ethics the test 
of truth is no longer the opinions of the many and the wise, 
but is, as in science, the perceived attributes and relations of 

The same assumptions, that moral words denote charac- 
teristics and that truth in ethics is discovered by observation, 
occur in the writings of Professor Moore. "My business", he 
says, "is solely with that object or idea, which I hold, rightly 
or wrongly, that the word 'good' is generally used to stand 
for"^ Moore's argument against the Naturalistic Fallacy is 
accepted in outline by all the intuitionists and is the basis of 
their belief that traditional moral philosophy rests on a mis- 
take. It is too weir known to require exposition in detail ; but 
a brief sunmiary will help to bring out the logical points to 
which I wish to draw special attention. 

Moore distinguishes sharply between 'analytic' statements, 
which are statements about the meaning of words, and 'syn- 
thetic' statements, which are about the facts of the case. 
Hedonists, for example, must make up their minds whether 
they want to say that 'good' just means 'pleasant' or to say 
that in fact pleasure alone is good. He then refutes each of 
these, alternatives and shows how hedonism derives its plausi- 
bility from a confusion of the two. Besides the distinction 
between analytic and synthetic statements Moore's logical 
apparatus contains the following points: 

I. Principia Ethica, p. 6 (my italics). The full argument against the 
Naturalistic Fallacy is given in chapters i and 2. 



(a) Words 'stand for' objects or ideas, and these are either 
simple or complex. 

(b) Definition is the analysis of a complex idea into the 
parts of which it is composed. Thus a chimaera is an animal 
with a lioness's head and body, with a goat's head growing 
from the middle of its back and with a snake in place of a tail. 
Similarly, when a word stands for a complex idea, explaining 
its meaning is a matter of enumerating and describing the 
simpler ideas of which the idea it stan'ls for is composed. 

(c) Words which stand for simple ideas cannot be defined, 
since the ideas for which they stand cannot be split into parts. 
'Yellow' is given as a typical example of such a word. Moore 
holds that 'good' also stands for a simple, and therefore in- 
definable idea; and the analogy with 'yellow' strongly sug- 
gests that we learn both what 'good' means and also what 
things are good by a process analogous to sense-perception. 

(d) Simple ideas are either 'natural' or 'non-natural'. In 
Principia Ethica this distinction is not expounded and we shall 
see that it gives rise to certain difficulties. Later, Moore was 
inclined to accept Broad's definition of a 'natural' object or 
characteristic as one which "either (a) we become aware of 
by inspecting our sense-data or introspecting our experiences, 
or (b) is definable wholely in terms of characteristics of the 
former kind together with the notions of cause and sub- 

The Naturalistic Fallacy consists in identifying certain 
ethical properties, such as Tightness and goodness, with 
natural properties, whether simple or complex, such as 
'pleasant', 'normal', 'more evolved', or 'according to God's 
will'. And the argument which refutes it is that, on reflection, 
we just see that the ethical properties are not identical with 
any of these natural ones. But in order to be able to see this 
we must first be able to see what goodness and rightness are. 
We do not literally see these properties with our eyes; but 
the faculty concerned is called 'non-sensuous intuition', 
I. The Philosophy of G. E. Moore, p. 62 and p. 592. 

B 33 


'awareness', 'apprehension', 'recognition', 'acquaintance' 
words which all strongly suggest an analogy with sight oi 

This analogy comes out well in the following passage: "Bu 
whoever v/ill attentively consider with himself zvhat is actually 
before his mind when he asked the question 'Is pleasure (oi 
whatever it may be) after all good ? ' can satisfy himself tha 
he is not merely wondering whether pleasure is pleasant. Anc' 
if he will try this experiment with each suggested definition ir: 
succession, he may become expert enough to recognize that ir 
every case he has before his mind a unique object. . . ."^ 

The same faculty that reveals to us the identity or non- 
identity of the objects for which words stand is also called intc 
play to decide questions of ethical fact, namely what things 
are good. Moore holds that, once the meaning of this ques- 
tion is clearly understood, the answer to it is just obvious: 
Thus the final touchstone turns out to be analogous to tha 
used in chemistry. Just as, in the last resort, we know tha 
the stuff in the bottle is sulphuric acid because it looks, smells 
and tastes like sulphuric acid; so, in ethics, we know hot? 
that one property is or is not identical with another and alsc 
what things have what properties by inspection. 

Professor Prichard represents the intuitionist thesis in its 
clearest and most uncompromising form. He uses the analog} 
with mathematics rather than that with empirical discourse 
but this makes little difference since he thinks of mathematica 
knowledge as an affair of 'seeing' or 'apprehending' and thuj 
appeals to the analogy with sight and touch at one remove 
It is in Prichard's writing also that we see most clearly ho^^ 
this analogy leads to the rejection of traditional moral philo- 
sophy as resting on a mistake. 

It is not a mistake in detail; but a mistake in principle. Il 
is that of supposing that when we are confronted by a dut} 
it makes sense to ask why we ought to perform it. Mora' 
philosophers have tried to answer this question; they have 
indeed made it the central question of ethics. But they have 

I. Principia Ethica, p. i6 (my italics). 2. Op. cit., p. 188. 



failed to find the answer because the question itself is a mis- 
taken or absurd one. It is absurd because our obligation to 
perform a duty is immediate and direct, neither requiring nor 
capable of supporting reasons. Our knowledge that we have 
a duty to perform is in the same way direct and immediate. If 
a man says that he does not see why he should do something 
that is in fact his duty we can only tell him to think again ; and 
'think' does not mean 'work it out'; it means 'look'.^ 

There are three points in Prichard's thesis to which I wish 
to draw attention, (a) Difficulties in the solution of a moral 
problem are all concerned with the non-moral facts of the 
situation. Once the "general and not moral" thinking has been 
properly done no further question can arise, since the recog- 
nition of the obligation is immediate and direct. There is 
really no such thing as moral thinking or wondering what I 
ought to do in this situation. We can only wonder whether 
the situation is really such as to give rise to a duty. 

(b) Although Prichard speaks mainly in terms of 'appre- 
ciating' a moral truth, he also speaks of 'recognizing' one; and 
although he avoids, perhaps deliberately, the word 'seeing', 
the whole tenor of his argument requires us to take 'appre- 
ciating' as analogous to 'seeing'. 

(c) Not only is moral knowledge the direct and immediate 
awareness of a datum ; we also know with a similar immediacy 
that we are knowing and not believing. Genuine knowledge 
(as opposed to belief) is infallible. A man who really knows 
cannot be mistaken ; and since he knows that he knows, he also 
knows that he cannot be mistaken. It is on this that Prichard's 
rejection of a criterion for either mathematical or moral 
knowledge rests. A criterion is both impossible and unneces- 
sary; impossible because it would be subject to the same 
sceptical doubts to which the knowledge it is used to test 
are subject, and unnecessary because the knowledge it is used 
to test is not subject to any genuine doubts at all. 


1 . Does Moral Philosophy rest on a Mistake ? Reprinted in Moral Obli- 
gation, pp. 7 and 15. 






In this and the following chapter I shall criticize the in- 
tuitionist theory; but my purpose is not to show that it is 
wholly mistaken or that its attacks on the Naturalistic Fallacy 
are misplaced. Rather I shall try to show that, on the negative 
side, the points which the intuitionist makes against the 
naturalist are correct : but that his way of making these points, 
the logical terminology in which he couches his arguments, 
misrepresents the very truths that the arguments are designed 
to bring out and makes his positive thesis in its own way as 
misleading as an account of moral discourse as were the earlier 
naturalistic theories. The intuitionist has noticed important 
differences between moral and empirical discourse; but he 
has marked these differences in the wrong way. 

The strength of intuitionism lies in its uncompromising in- 
sistence on the autonomy of morals. To put the point briefly 
and in my own way, practical discourse, of which moral dis- 
course is a part, cannot be identified with or reduced to any 
other kind of discourse. Ethical sentences are not, as Moore 
so clearly shows, psychological or metaphysical or theological 
sentences. Almost all earlier theories had tended to reduce 
ethical concepts and sentences to those of some other subject, 
usually psychology; they tried to define words such as 'good' 
and 'ought' in terms, for example, of the satisfaction of desire 
or of pleasure and pain. Against all such attempts the intui- 
tionists produce a crushing argument which is derived (sur- 
prisingly) from Hume. 

In a celebrated passage Hume says: "I cannot forbear 
adding to these reasonings an observation which may perhaps 
be found of some importance. In every system of morality 



which I have hitherto met with I have always remarked that 
the author proceeds for some time in the ordinary way of 
reasoning, and estabhshes the being of a God, or makes 
observations concerning affairs; when of a sudden I 
am surprised to find, that instead of the usual copulations of 
propositions, is and is not, I meet with no proposition that is 
not connected with an ought, or an ought not. This change is 
imperceptible ; but is, however, of the last consequence. For 
as this ought or ought not expresses some new relation or 
affirmation, it is necessary that it should be observed and 
explained ; and at the same time that a reason should be given 
for what seems altogether inconceivable, how this new relation 
can be a deduction from others that are entirely different 
from it."i 

Freely translated into modern terminology, what Hume 
means is this. In ail systems of morality we start with certain 
statements of fact that are not judgements of value or com- 
mands; they contain no moral words. They are usually state- 
ments about God or about human nature, that is to say about 
what men are and in fact do. We are then told that because 
these things are so we ought to act in such and such a way; 
the answers to practical questions are deduced or in some 
other way derived from statements about what is the case. 
This must be illegitimate reasoning, since the conclusion of an 
argument can contain nothing which is not in the premises, 
and there are no ' oughts ' in the premises. 

We are not concerned with the validity of this argument 
as a criticism of traditional ethics. It might be argued that 
Aristotle and certain Christian writers did not make this 
mistake, since their premises were really disguised value- 
judgements or 'ought '-judgements from the start. But some 
philosophers, notably the hedonists, certainly did make this 
mistake, and of them Hum.e's criticism is an unanswerable 

I. Treatise, Book III, Part i, Section i. 

3. Cf . Bishop Mortimer : Christian Ethics, p. 7. " The first foundation 
is the doctrine of God the Creator. God made us and all the world. 



In their various ways the writers of the intuitionist schoo 
have but repeated Hume's argument. I may know that a cer- 
tain action will please God or maximize my own pleasure oi 
produce the greatest happiness of the greatest number; bui 
this is all knowledge of what is or will be the case. It stil' 
makes sense to ask whether I ought to do the action. As 
Prichard points out, a 'link' is required to connect the state- 
ments of fact with an injunction to do or not do something or. 
in my terminology, to connect the answer to a theoretical 
question with the answer to a practical one. We must now see 
whether the intuitionist is in a better position to provide this 
link, to bridge the gap between 'is' and 'ought' than his 
naturalistic predecessors. 

At first sight it seems that he is; for he has so arranged 
matters that no gap exists. Earlier moralists tried to derive 
obligation-statements from statements of other kinds; but. for 
an intuitionist, obligations are immediately and underivatively 
known and require no deduction. The demand for a bridge, 
for an argument connecting 'ought' to 'is' is senseless because 
we are directly confronted by 'oughts'. But a closer examina- 
tion will show that this way out of the difficulty is a spurious 
one. I shall argue later that the intuitionist cannot both main- 
tain the immediate and underivative character of moral know- 
ledge and also the analogy with empirical discourse which 
justifies his use of such terms as 'see', 'recognize', 'true', 
'mistaken', 'know', 'fact', and 'objective'.^ In this chapter I 
shall try to show that his way of representing moral knowledge 
as theoretical knowledge still leaves him with a gap to be 

Because of that He has an absolute claim on our obedience. We do not 
exist in our own right, but only as His creatures, who ought therefore to 
do and be what He desires." This argument requires the premise that a 
creature ought to obey his creator, which is itself a moral judgement. 
So that Christian ethics is not founded solely on the doctrine that God 
created us. 

I. Chapter 4, below. 



Moral knowledge is represented by intuitionists as knowledge 
that a certain object has a certain characteristic. To learn a 
moral truth is like learning that Henry VIII had six wives or 
that a Centauri is 4I light years away. The difference between 
moral characteristics and those that we learn about in science 
and in history is marked by calling them non-natural. But 
moral judgements are treated as descriptions of features of 
the universe, and the fact that these features are so peculiar 
as to merit the epithet 'non-natural' in no way affects the 
status of moral judgements as descriptions. 

Yet in spite of this the core of the intuitionist argument 
against naturalism is that normative or evaluative propositions 
cannot be deduced from descriptive ones. Intuitionists recog- 
nize the importance of moral approval and disapproval and 
of moral sentiments such as remorse ; and it might seem that a 
philosopher who was prepared to assert that there are special, 
non-natural, moral emotions is not guilty of the naturalistic 
fallacy. At least he would not be guilty of trying to reduce 
something specifically moral to some non-moral thing. But this 
concession to the non-natural status of moral concepts is not 
enough to satisfy the intuitionist. In criticizing ' approval ' and 
'emotive' theories of ethics Ross says: "For as of moral 
approval, so of the emotion of obligation, we must say that 
it is not a blind feeling that arises in us, we know not why, 
on contemplating a possible act. It is an intellectual emotion 
which arises only when we judge the act to have a certain 
character, say that of producing a maximum of good, and to 
be on that account obligatory."^ Moral emotions could not 
occur unless we were aware of moral facts. 

The intuitionist's answer to the question 'why should I be 
moral?' - unless, like Prichard, he rejects it as a senseless 
question - is that, if you reflect carefully, you will notice that 
a certain act has two characteristics, (a) that of being obliga- 
tory and (b) that of producing a maximum of good or of 
I. Op. cit., p. 26. cf. p. 23. 



being a fulfilment of a promise or the payment of a debt, etc. 
There are a certain number of these 'right-making' charac- 
teristics and they are related to obligation in the following 
way. An act which has one of the right-making characteristics 
would always be right; but these characteristics might con- 
flict and we cannot be obliged to do conflicting actions; so an 
action is obligatory only when the balance of rightness is on 
its side. You decide which action has the balance of rightness 
on its side by estimating the stringency of the various 'claims' 
or 'prima facie rightnesses' concerned. 

Moreover, if you have noticed these characteristics you will 
feel a special moral emotion of obligation; and you will not 
feel this special emotion if you have not noticed the charac- 

But suppose all this has taken place. I have noticed the right- 
making characteristics and the rightness ; and I feel the emotion, 
of obligation. Does it follow that I ought to do the action to- 
wards which I feel the emotion ? If Hume's argument is valid 
at all, is it not equally valid against this deduction ? It cannot 
be evaded by merely calling the characteristic and the emotion 
'non-natural'; copious use of this epithet serves only to dis- 
guise Hume's gap, not to bridge it. 

In representing moral knowledge as theoretical, an affair of 
being aware of or noticing 'phenomena', albeit of a very special 
kind, the intuitionist is drawing our attention to an analogy 
between ethics and empirical science. We learn some things, 
he says, by hispecting our sense-data, others by introspecting 
our experiences; and a third world, a world of non-natural 
characteristics, is revealed to us by a third faculty called 'intu- 
ition'. (Perhaps there is a fourth world in which the Beautiful 
and the Ugly reside, which is revealed by a fourth faculty. In- 
tuitionists are divided on this point, some claiming that there 
is, others that there is not. Surely this in itself is suspicious.) 

I. It is not quite clear what the connexion between noticing the 
characteristics and feeling the emotion is supposed to be. Is it logically 
or casually impossible for a man to feel just this emotion if he has not 
noticed the characteristics ? Or is it impossible in some other way ? 



It is only in this way that moral knowledge can be repre- 
sented as theoretical. For what is it to call something 'theor- 
etical ' but to drav/ an analogy between it and something that 
is theoretical in a straightforward sense, such as empirical 
science, and to treat the logical terminology that we use when 
talking about the latter as equally applicable to the former? 
But, if the analogy is just, are we not simply confronted with a 
new set of data or phenomena or characteristics? And from 
statements to the effect that these exist no conclusions follow 
about what I ought to do. A new world is revealed for our in- 
spection; it contains such and such objects, phenomena, and 
characteristics ; it is mapped and described in elaborate detail. 
No doubt it is all very interesting. If I happen to have a thirst 
for knowledge, I shall read on to satisfy my curiosity, much as 
I should read about new discoveries in astronomy or geo- 
graphy. Learning about 'values' or 'duties' might well be as 
exciting as learning about spiral nebulae or waterspouts. But 
what if I am not interested ? Why should I do anything about 
these newly-revealed objects ? Some things, I have now learnt, 
are right and others wrong; but why should Ido what is right 
and eschew what is wrong ? 

"But", you will say, "all this is monstrous. It is regrettable, 
perhaps, that you are not interested in nebulae or waterspouts, 
- regrettable, but not blameworthy. But not to be interested in 
the difference between right and wrong ! Not to be able to see 
the difference, not to feel the obligation to do the one and 
eschew the other! This is not regrettable; it is v/icked, im- 
moral, sinful, inhuman. And anyhow I just don't believe you. 
You are propounding a philosopher's paradox, playing with 

And of course you are right. Of course the question, 'Why 
should I do what I see to be right ? ' is not just an immoral ques- 
tion, but an absurd one. In ordinary life we should be puzzled 
by a man who said "Yes, you have convinced me that it is the 
right thing to do; but ought I to do it?". We shouldn't know 
how to answer him, not because we could not think up any new 
arguments, but because, in conceding that it is the right thing 



to do, he has aheady conceded that he ought to do it. Our 
puzzle is to understand the distinction he seems to be making 
between what it is right for him to do and what he ought to 
do. He must be able to see a distinction here, since he grants 
that something is right but still wonders whether he ought to 
do it. He has really solved his practical problem; but he seems 
to think that there is an extra step to be taken. What could 
this step possibly be ? 

But the argument I have used is not designed to convince 
the ordinary man that there is really no difference between 
right and wrong or to prove to him that he has no obligations 
or even to insinuate doubts about the propriety of passing from 
'this is right' to 'I ought to do this'. It is a philosophical 
argument designed to refute a special, philosophical account of 
what obligation and our knowledge of obligations are. 

These questions, which are absurd when words are used in 
the ordinary way, would not be absurd if moral words were 
used in the way that intuitionists suppose. For in ordinary life 
there is no gap between 'this is the right thing for me to do' 
and 'I ought to do this' into which a sceptic might insert a 
wedge. But if 'X is right' and 'X is obligatory' are construed 
as statements to the effect that X has the non-natural charac- 
teristic of rightness or obligatoriness, which we just ' see ' to be 
present, it would seem that we can no more deduce ' I ought to 
do X' from these premises than we could deduce it from 'X is 
pleasant' or *X is in accordance with God's will'. A gap of 
which ordinary language knows nothing has been created be- 
tween 'X is obligatory on me' and ' I ought to do it' ; and this 
gap requires to be bridged. 

To escape from this argument the intuitionist might reply 
that his talk about recognizing non -natural qualities in actions 
and things has been taken far too seriously. There is really, he 
will say, no such characteristic as 'obligatoriness'. Saying that 
an action is obligatory on me is a convenient but strictly an 
incorrect way of saying that I ought to do the action. 

This line of argument does indeed eliminate the unbridge- 
able gap between what is obligatory and what I ought to do, 



but it does so only by conceding that the characteristic of ' obli- 
gatoriness ' is a myth. And he might deal with 'rightness' in 
the same way - and at the same cost. He might say that calling 
an action ^ prima facie right ' is a convenient, but strictly an in- 
correct way of saying that I ought to do it if there is nothing 
that I ought to do more. But he still has on his hands the task 
of explaining the fact that no action is merely right, but always 
'right as being of a certain character', for example as being 
the payment of a debt or the fulfilment of a promise. 

And what, meanwhile, has happened to the intuitionist 
theory ? The account of obligation as a non-natural character- 
istic of actions was designed both to tell us what obligation is 
(not to tell us what our obligations are) and to explain how 
we know what we ought to do. It fails in both tasks; and for 
the same reason. For it turns out to be, not an explanation, 
but a restatement in technically convenient, but strictly incor- 
rect language of the facts to be explained. We are no nearer to 
knowing what obligation is, because it is now conceded that 'X 
has the characteristic of obligatoriness ' is just another way of 
saying 'I ought to do X'. And the intuitionist' s reply to the 
question ' How do I know what I ought to do ? ' is equally un- 
enlightening. For it turns out to be: "You know what you 
ought to do by intuiting the non-natural characteristic of obli- 
gatoriness that inheres in certain actions. But, my dear sir, do 
not be alarmed by this mysterious phrase. It is only another 
way of saying that you know that you ought to do those 
actions." We know what we ought to do by knowing what we 
ought to do. Opium sends us to sleep because it has a virtus 


The authors so far considered would not claim that their talk 
about objective, non-natural properties was intended to help us 
to solve practical problems. They claimed to show how the 
words in our moral vocabulary, 'good', 'right', 'obligatory', 
and ' ought ' are to be construed. What they offer us is in effect 



a translation of ordinary moral sentences into a new termin- 
ology; And such a translation might well be enlightening. We 
might, for example, wonder whether moral judgements were 
more like statements of empirical fact or like expressions of 
taste; and, by showing that they can be translated into the 
language of objective properties, by making out a case for the 
analogy with empirical discourse, they might settle this old 
dispute in favour of Reason, not Sentiment. 

But how would we discover whether such a translation can 
profitably be made ? The new way of saying what a moral de- 
cision or conflict or ju4gement is ought to make these notions 
clear. I shall now try to show that, even if the objective- 
property theory is correct, it throws no light on one of the 
most important and difficult problems of ethics which it ought 
to beable to elucidate. 


For the extreme intuitionist there is no problem about recon- 
ciling the conflicting moral insights of diff^erent people. If I 
clearly apprehend the truth I may grieve at your inability to 
see it; but I cannot be troubled by an anxiety that I may have 
failed to see it myself. But if once this possibility is admitted, if 
I can wonder whether what I took to be a genuine intuition 
might not have been a spurious one, the problem becomes 
serious. How does the appeal to intuition help us to settle 
radical diff^erences that cannot be attributed to disagreement 
about the non-moral facts of the case? 

Let us see first of all how a subjectivist would describe such 
disagreement. If I disagree with you about the Tightness of an 
action or the goodness of a person or the morality of some 
general line of conduct, my arguments, he would say, must be 
of one or two kinds. 

(a) I might try to convince you that you were mistaken 
about some non-moral fact. For example you thought it wrong 
of me to have refused money to that blind beggar, and the dis- 
pute ceases when I point out that I happen to know him to be 



a fraud. There was never any moral dispute between us, since 
we both wished to advocate charity towards genuine beggars, 
not frauds. 

(b) I might try to convince you that my moral judgement 
follows from or is a special case of some more general moral 
judgement which I know that you accept. For example you 
might approve of giving money to beggars because it relieves 
distress - and we both approve of relieving distress. If I con- 
vinced you that your sort of charity creates more distress than 
it alleviates, which is a question of empirical fact, then you 
must withdraw your censure and the dispute ceases. 

Both these form.s of argument presuppose that there is some 
common moral ground between us, if only we can find it. 
Many disputes, say the sub] activists, turn out to be disputes 
about the facts or about the validity of deductions made from 
those facts and our common ethical premises. But, they say, a 
point may be reached when there is no disagreement of a 
factual or logical kind ; and yet a moral disagreement remains. 
You approve of one thing and I approve of another. And when 
this point is reached it is tautological to say that the dispute 
cannot be settled by rational argument. We must either agree 
to differ or resort to flattery, cajolery, bribery, or other non- 
rational methods of persuasion. Or I must use force or submit 
to your using force on me. 

Now this is not a palatable conclusion, especially when it is 
erroneously believed to lead to the doctrine that Might is 
Riglit.^ And one of the most important claims that has been 
made for 'ol)jective values' is that they enable us to escape 
from it. We do not think it necessary to resort to force where 
logical, mathematical, or empirical truths are concerned. And 
if moral truths can be apprehended in the same sort of way, if 

I. The conclusion that Might is Right does not in fact follow from 
the subjective thesis, though it is surprising how often it has been thought 
to do so. Subjectivism is a theory about the nature of moral judge- 
ments. To put it crudely, it is the theory that they are expressions of 
a man's personal attitudes, his approvals and disapprovals. It does not 
imply the theory that most men approve of trampling on their neigh- 
bours; nor need the subjectivist himself approve of this. 



there are criteria that anyone can use to discover them, it ought 
to be possible to settle moral disputes without recourse to 
force, by rational argument. 

But does the notion of objective values in fact help us to 
reconcile conflicts without recourse to force ? Clearly it cannot. 
For I will claim this status for the values that I intuit, and you 
will claim it for yours. If the empirical analogy is used, it will 
make sense for me to say that my view is 'true' and yours 
'mistaken', terms which the subjectivist must deny himself; 
but we are no nearer a solution of the conflict, which still 
remains exactly what it was before, a conflict between the 
moral facts that I claim to descry and those that you claim to 
descry. If I do not recognize the truth of your statements of 
objective moral fact, you can do nothing to convince me by 
argument but are, like the subjectivist, thrown back on force 
or acquiescence. Wars are not fought over logical or empirical 
issues, just because men do in the end agree about what they 
see to be the case or ' see ' to be logically valid. But wars are 
fought over moral issues, and the only difference between the 
subjectivist and the objectivist is that to the former this fact is 
not surprising. 

Theoretically the objective theory cannot help us to recon- 
cile conflicts, since it conceals a difficulty endemic in all theories 
involving 'intuition'. Intuitions of objective properties are 
either infallible or they are not. If they are fallible, the mere 
existence of an objective property or value is no guarantee that 
anyone has apprehended it properly. However convinced you 
may be that you are right, it is still open to me to deny the 
genuineness of your intuition. If, on the other hand, intuitions 
are infallible, then disputes cannot be genuine. If I disagree 
with you, you must charge me either with insincerity or with 
moral blindness. And that this account of the matter is false is 
shown by the fact that we do often allow others to be sincere 
when their moral views differ from our own. 

And in practice the objectivist is, as we should expect, in a 
far worse position for solving moral conflicts. He necessarily 
attributes his opponent's denial of the truth to wilful perver- 



sity; and, holding as he does that in spite of his denials his 
opponent must really see the truth all the time, he realizes that 
what his opponent needs is not argument but castigation. For 
arguments cannot convince a man who already sees the light. 
The objective theory, so far from minimizing the use of force 
to settle moral conflicts, can be, and constantly has been used 
to justify it. It is no accident that religious persecutions are the 
monopoly of objective theorists. 

The theory underlying persecution is admirably explained 
by Samuel Clarke. " These things are so notoriously plain and 
self-evident, that nothing but the extremist stupidity of mind, 
corruption of manners or perversity of spirit can possibly make 
any man entertain the least doubt concerning them. For a man 
endued with Reason to deny the truth of these things, is the 
very same thing, as if a man who had the use of his sight should, 
at the same time that he beholds the sun, deny that there is any 
such thing as light in the world; or as if a man that under- 
stands geometry or arithmetic, should deny the most obvious 
and known proportions of lines or numbers, and perversely con- 
tend that the whole is not equal to all its parts. . . . And 'tis as 
absurd and blameworthy to mistake negligently plain right and 
wrong ... as it would be absurd and ridiculous for a man in 
arithmetical matters ignorantly to believe that twice two is not 
equal to four or wilfully and obstinately to contend, against his 
own clear knowledge, that the whole is not equal to all its 

If the analogy between moral knowledge and sight or 
mathematics is as close as Clarke makes out, those who disa- 
gree vdth us on a moral issue must be insincere. It is force, not 
argument that they need. It is hardly necessary to add that 
this theory has had the most tragic consequences in inter- 
national affairs. To suppose that people whose professed moral 
principles differ from ours do not really hold them is to invite 

I. On Natural Religion. Selby-Bigge: British Moralists, Vol. II, pp. 
6 and 13 (my italics). 



The Analogy between Ethics and Science 


How are we to interpret this talk of a special ' world ' of' order ' 
of values which man's moral consciousness apprehends or of 
special 'qualities' (like yellow, yet so unlike yellow), that we 
are aware of, but not by means of the senses ? We have seen 
that if we take it quite literally ethics becomes the geography 
of a special world out of space and time and that moral know- 
ledge, in being represented as knowledge of what is the case, 
becomes cut off from practical loiowledge, from doing, decid- 
ing, choosing, advising, and exhorting in such a way that 
special intuitions are required to fill the gap. And if we take it 
less literally it turns out that all these mysterious phrases are 
only a rather pretentious way of saying - what we all know- 
that some things are right and others wrong, whereas they 
were intended to explain how we know this. 

The intuitionist thesis cannot really be taken in this straight- 
forward way and it is obviously unfair to take it so. The intu- 
itionist is not drawing our attention to the obvious fact that we 
have moral experience; he is inviting us to construe this ex- 
perience in a special way. To be specific, he is inviting us to 
construe this experience as being analogous to seeing. Moral 
judgements are reports of what we see when we look at the 
non-natural world in the same sort of way that empirical state- 
ments are reports of what we see when we look at the natural 
world. This point comes out clearly in the passage quoted from 
Clarke. But much has happened in philosophy since Clarke's 
day and the analogy must now be put more warily. We must 
guard against the accusation of naturalism at every step by 
insisting that the moral world and all that is in it is 'non- 
natural'. This will remind us that moral qualities are not 



ordinary qualities, moral judgements not ordinary statements 
and moral insight not eyesight. 

I have already suggested that this analogy fails to explain at 
least one of the most important problems of ethics, the problem 
of reconciling moral conflicts; and I shall now try to show that 
the analogy is mistaken in principle. It is not that the analogy 
is wholly false. The very fact that we use objective terminology 
in moral matters, that we say 'this is good' rather than 'I 
approve of this ', that we call moral judgements true or false, 
shows that there is some analogy between moral and empirical 
discourse. I shall try to shov/ that the analogy breaks down at 
a crucial point; but it does not follow, as some philosophers 
have thought, that we must then say that moral judgements 
are mere matters of taste of what I happen to like. 

To believe this is to take a mistaken view of the part played 
by analogies in philosophy. Suppose that a man were told that 
a piece of music was either by Haydn or by Mozart and asked 
to guess which. He might notice many Haydnesque and many 
Mozartesque features and he would decide in favour of the 
composer to whose other works he found the best analogies. 
He uses these analogies to decide which answer is correct, 
knowing that the composer must be one or the other and not 

But philosophical analogies are not like this. We find that 
moral judgements are in some v;ays like empirical statements 
and in other ways like expressions of taste ; but it is not incum- 
bent on us to say that moral judgements must he the one or the 
other. In fact they are obviously neither. We use analogies, not 
to help us to decide which they are, but to elucidate what a 
moral judgement is by showing in what ways it is like an em- 
pirical statement and in what v/ays it is like an expression of 
personal taste. 

The analogy with empirical discourse is designed to bring 
out the fact that moral judgements are like empirical state- 
ments in the following ways. They describe or refer to facts. 
Facts are discovered, not created by thinking. If the facts are 
such and such anyone who believes them to be so is correct and 



his belief is true ; anyone who believes them to be something 
different is mistaken and his belief is false. It is a cardinal point 
that the truth or falsity of an opinion depends wholly on what 
the facts are, not on whether anyone holds it. However un- 
likely it might be, it is always possible for everyone to be mis- 
taken on a certain point of fact. This is the sort of language 
that we must be prepared to use if we call something ' an ob- 
jective matter'; and the thesis that moral judgements are 
objective is, in effect, the thesis that this sort of language can 
properly be used about them. 

By the test of ordinary usage objective language is, as I 
have said, appropriate to moral discourse. But it is in empirical 
discourse that these objective expressions have their most 
typical standard application and, when we apply them else- 
where, we are apt to have empirical discourse in mind. And we 
must now see whether v/e can find in moral discourse just those 
features that lead us to use objective language in empirical 

We have seen that the intuitionist represents moral know- 
ledge as an affair of immediate insight into, or apprehension 
of a ' datum ' that is directly presented to us. This might do as 
an account of moral knowledge in a case in which we can have 
no serious doubts. There is point in Clarke's analogy between 
a man who denies a plain moral truth and a man who, having 
the use of his sight, denies that he can see the light. It is in 
cases where there is doubt about the moral ' facts ' that we are 
involved in difficulties. 

It is curious how little attention has been paid by intuition- 
ists to the problem of conflicting obligations, which seems to 
be one of the most important in ethics, except by way of criti- 
cism of those writers who have thought to solve it in terms of 
the intensity or quantity of pleasure which different actions 
would bring about. Ross says of it : " In this region our know- 
ledge is very limited. While we kfiozv certain types of action to 
be prima facie obligatory, we have only opinion about the 
degree of their obligatoriness. . . . While we can see with cer- 
tainty that the claims exist, it becomes a matter of individual 



and fallible judgement to say which claim is in the circum- 
stances the overriding one. In many such situations, equally 
good men would form conflicting judgements as to what their 
duty is. They cannot all be right, but it is often impossible to 
say which is right; each person must judge according to his 
individual sense of the comparative strength of various 

Broad speaks of " allowing due weight to the relative urgency 
of each claim " and he also compares " the claims which arise 
from various right-tending and wrong-tending characteristics 
to forces of various magnitudes and directions acting on a body 
at the same time."^ The empirical analogy has seldom been so 
clearly put. Deciding whether I ought to do this or that in a 
case where I am confronted by conflicting obligations is like 
estimating the magnitude of a force or like estimating the rela- 
tive weights or temperatures of two different bodies. I shall 
try to show that the analogy between estimating the stringency 
or urgency of various moral claims and estimating the magni- 
tude or intensity of an empirical property is fundamentally 
misleading, and that we cannot both maintain this analogy and 
maintain that "each person must judge according to his in- 
dividual sense of the comparative strength of the claims". The 
concession to subjectivism which is, as we shall see, involved 
in representing moral experience as the awareness of a datum 
is fatal to the attempt to treat moral qualities or 'phenomena' 
as objective. 


It is a truism that we detect the presence and estimate the 
degree of empirical properties 'by means of the senses', by 
looking, feeling, tasting, and smelling; since 'empirical proper- 
ties ' are just those properties of which the presence and degree 
are discovered in this way. The language of 'awareness', 
* perceiving ', ' observing ', and ' apprehending ' has its standard 
I. Op. cit., pp. 188-9 (author's italics). 2. Op. cit., p. 553. 



application here and cannot, therefore, significantly be called in 
question. Many empirical properties are not directly perceived; 
but even though estimating their degree involves indirect 
methods, the appeal must, in the end, be to observations.^ 

The logic of the language that we use in empirical discourse 
reflects the conditions under which such language is used, and 
it is these conditions that make possible and give point to the 
objective-subjective distinction and all that it implies. We must 
rely on observation; but we also know that our senses some- 
times deceive us. Direct observation often gives different 
results when there is no reason to suppose that the object ob- 
served has changed* The suitcase that felt light when I left the 
station feels heavy after I have walked a mile; yet no one has 
put anything in it. In one of the standard examples of optical 
illusion a certain line appears to be curved but when a ruler is 
put against it it appears to be straight; yet no one has altered 
the position of the ink on the paper. A piece of paper that 
looked white before looks red under a red light; yet no one 
has painted it. 

We could cope with these situations without the objective- 
subjective contrast. We could say that the suitcase was light 
and is now heavy, the line was curved and is now straight, the 
paper was white and is now red. But this way of putting things 
has an obvious disadvantage; it makes all our talk about the 
properties of things dependent on the conditions of observation 
and, above all, on the position and condition of the observer's 
body. And in fact we adopt a different method. We make use 
of a double language in which ' is ' is contrasted with ' looks ' 
and 'feels'. 

The double language operates in two ways, (a) In some 
cases, for example that of colour- words, the test of whether a 
thing really has a property or not is whether it appears to a 
normal observer to have it under certain standard conditions ; 
thus the test for whether something is red is whether it appears 

I. The subject of empirical estimating and the contrast between 
' directly-perceived' and ' scientific ' properties is more fully discussed 
by S. E. Toulmin: The Place of Reason in Ethics, chapters 7 and 8. 



to be red to a normal observer in ordinary sunlight ; and it is 
important to notice that both the normal observer and the 
standard conditions can be defined without circularity. 

(b) But in many cases, especially where scientific properties 
are concerned, the test is more indirect. 'This feels heavier 
than that' corresponds to 'this looks red'; but we do not 
decide the real relative weights of things by reference to the 
sense of muscular strain of a normal person in standard con- 
ditions. We appeal, as in many other cases, to the sense of 
sight. We put the objects on a pair of scales and see which scale 
is depressed or put them in turn on a spring balance and see 
how much the spring is stretched. Similarly we test the ' real ' 
temperature of a bowl of water, not by putting our hand into 
it, but by seeing the height to which the mercury in a ther- 
mometer immersed in it rises. 

Now it is an essential feature of the double language of 
* looks or feels ' and ' really is ' that, while the observer himself 
is allowed to be the best judge of how a thing looks or feels to 
him, he is not allowed to be the best judge of what it really is. 
If two cakes feel equally heavy to me but the balance shows 
one to be heavier than the other, I must say that one of them 
was really heavier all the time ; but I am not compelled to say 
that it must have felt heavier. Obviously it did not. There is 
something about the first observation that I stubbornly wish 
to retain, since to abandon it would be to cast doubt on all 
observation; and it is a truism that we must rely on observa- 
tion in the end. But I also want to admit that I was mistaken, 
since the balance shows that the cakes are not equal in weight 
and I have no reason to believe that anything has happened to 
either of them. The double language enables me to deal with 
this sort of situation. " They hoth felt the same to me, " I shall 
say, "but one of them must really have been heavier all along." 

There is no need to investigate the philosophical problem 
whether looks and feels are absolutely incorrigible, whether 
there are what some philosophers call ' bare sense-data ' in re- 
porting which I cannot be mistaken. It is enough to notice that 
the language of ' looks ' and ' feels ' is relatively incorrigible, a 



language into which we can retreat if we do not want to com- 
mit ourselves to the hazards of saying what a thing really is. 
And I am safer here than in the ' is ' language just because I am 
the best judge of how a thing looks or feels to me, though not 
of its real properties. 

But there is an important condition that must be satisfied if 
any particular test is to be used as a test for a ' real property '. 
There must be a very wide measure of agreement in its appli- 
cation. If balances and thermometers did not give a greater 
measure of agreement than the felt heft or warmth of things, 
we could not appeal to them from feeling in the way that we 
in fact do. Now it is an empirical fact that there are indirect 
tests, loosely correlated with felt weight and temperature, 
which are such that their application gives more accurate, more 
stable, and better agreed results than does the application of 
the direct tests. But it is a logical fact that nothing could be 
used as an indirect test unless it fulfilled these conditions. If, 
for example, people disagreed as much about the readings of 
thermometers as they do about felt temperature and if the 
thermometer reading were liable to fluctuate when there was 
no reason to suppose that any change had occurred in the ob- 
ject, we could not (logically) use thermometers as tests for 
real temperature. And if no devices had been discovered which 
had these logical properties of loose correlation with feeling 
and a high degree of stability and agreement we could not 
(logically) treat temperature as a real property at all. ' Hot' and 
'cold' would be wholly subjective words. 

If these conditions were not fulfilled we should have, in 
short, no use for the double language of ' feels ' and ' is '. We are 
able to contrast real weight with felt weight, not because we 
have some mysterious insight into an unobservable real prop- 
erty that somehow underlies or causes the felt and seen weights 
of things, but because there is a marked contrast between the 
balance test and the feeling test. There is no more need to 
claim universal agreement in the application of tests for real 
properties than there was to claim absolute incorrigibility for 
looks and feels. Some people are notoriously bad experimen- 



ters ; but this fact does not tempt us to deny that there are real 
properties. It is enough that the contrast should be marked. 

It should be noticed that I am not arguing that general 
agreement is a conclusive test of the presence or degree of a 
real property. Our ordinary objective language makes truth 
independent of opinion and allows for the possibility, however 
remote, of everybody being mistaken. A Gallup Poll could not 
settle the question whether the earth is round. General agree- 
ment is not a test of truth, but it is a necessary condition of the 
use of objective language. We could not treat roundness as 
an objective property, we could not talk about things being 
really round or say that statements about roundness were ob- 
jectively true or false, unless two conditions were fulfilled: 
(a) We must agree about the tests for roundness, what ob- 
servations would strengthen and what would weaken the case 
for the earth being round, and (b) the tests used must be such 
as to give a high degree of agreement in their application over 
a wide field. It is not necessary that there should be agreement 
in a particular case or that the agreed opinion should be cor- 
rect. If everyone agreed that the thermometer reading was 
100° F., everyone could (logically) be mistaken ; but it would 
be impossible to speak of people being correct or mistaken in 
their readings of thermometers or of statements about these 
readings being true or false if such disagreement were 

This last point has an important bearing on the relevance of 
the existence of moral disagreement to the objective-subjective 
controversy. Subjectivists point to the fact of disagreement as 
conclusive proof that moral properties are not objective. To 
this it is replied that the conclusion does not follow because 
people often disagree where objective, empirical properties are 
concerned. Moreover there is no reason to suppose that there 
ought to be as much agreement in the reports of moral intu- 
ition as there is in the reports of what we see. Moral intuition 
might be a very imperfectly developed faculty and better de- 
veloped in some men than in others. But this reply misses the 
point of the objection. If people really do disagree much more 



widely and violently in their reports of moral * data ' than they 
do about empirical data, the conclusion must be, not that moral 
opinions are subjective or that one party must be mistaken, but 
that the language of 'correct' and 'mistaken', which is an 
essential part of the objective terminology, cannot be applied. 
For there are no tests for deciding which party is correct and 
which mistaken. 

If we are to justify the analogy between moral properties 
and empirical properties, which is implied by the use of objec- 
tive terminology, we must show that there is a contrast in 
moral matters between 'is right' and 'seems right', which 
corresponds to the contrast between ' is red ' and ' looks red ' or 
betvv^een 'is heavier' and 'feels heavier'. But this is exactly 
what the intuitionist cannot do ; for in making direct awareness 
the test of real ethical properties he eliminates the whole point 
of the objective-subjective contrast. It is not that ethical prop- 
erties are subjective or even that the objective-subjective con- 
trast is wholly misplaced in ethics, but that the use of this 
contrast is quite incompatible with intuitionism. 


From their talk about estimating the strengths of claims or the 
stringency of obligations and from Broad's comparison of these 
properties to forces in mechanics it would seem that intuition- 
ists have in mind mainly the analogy with type (b) properties. 
But this can hardly be the case. For in type (b) cases the real 
and the apparent property are detected by different senses ; and 
no one supposes that ' apparent stringency ' and ' apparent moral 
value' are revealed to one moral sense while their 'real' 
counterparts are revealed to another. It is therefore to type (a) 
properties that the analogy is drawn; and I shall now try to 
bring out an important difference between recognizing a moral 
property and discovering what colour a thing really has. 

We have seen that the double language only has point if we 
allow that the observer himself is the best judge of looks and 



feels but not of what a thing really is. And this enables us to 
admit ourselves consistently wrong in empirical cases. If a 
man finds that his judgements about colour differ consistently 
from those of others, he will admit himself to be colour-blind. 
He might start by saying that two things were the same 
colour; but if he finds that everyone else says that they are 
different he will retreat into the language of ' looks to me '. In 
the same way a man who finds that his readings of scales and 
meters differ from other people's does not immediately write 
to Nature to claim that his observations upset some well-known 
iscientific law; he recognizes that he is a bad experimenter and 
probably takes up some other career. The existence of colour- 
blind persons and bad experimenters does not prevent our 
using objective language; for the dissentients are willing to 
lallow that the common opinion is correct, however much it 
!may conflict with their own experience. If each man had to 
judge the real colour of an object by his individual sense of 
icolour, the very distinction between ' is red ' and ' looks red to 
ime' would have broken down and we should have no use for 
Iredness as a real property at all. For, where real properties are 
Iconcerned, general agreement is admitted, even by a dissentient, 
to be the criterion for the property. 

But, according to the intuitionist, immediate insight is the 
test of the presence and of the degree of stringency of an obli- 
gation. " Each person must judge according to his individual 
sense of the comparative strength of the various claims." It is 
admitted that, at least as far as questions of stringency are 
concerned, " equally good men would form conflicting judge- 
ments". But this admission is fatal to the empirical analogy. 
For, in the first place, if the appeal is to each person's indi- 
vidual sense, it is impossible to know whether those who dis- 
{agree are really "good men". We cannot, as we can in empiri- 
cal cases, define the normal or standard observer without 

j Secondly, even if we have somehow been able to select our 
panel of experts, the disagreement that arises between them is 
of a kind that could not possibly occur in the case of objective 



empirical properties. No doubt the greatest experts can dis- 
agree about theories, whether in science or elsewhere; but 
theories are not ' data ' or reports of observation. The disagree- 
ment which Ross has in mind is not that between rival moral 
theorists, for example Hobbists, Kantians, and Utilitarians, 
but disagreement about something which is represented as 
analogous, not to scientific theories, but to reports of observa- 
tions. The parallel case would be that in which a number of 
scientists failed to agree about the reading of a scale or a meter 
or about the colour of an object. And we have seen that, if this 
sort of disagreement were of frequent occurrence, the property 
in question could not be treated as a 'real' or 'objective' 
property at all. 

If there really is widespread disagreement about the relative 
stringency of two claims, it is logically impossible that strin- 
gency should be a real property analogous to the real properties 
of empirical discourse. It is not just that we cannot say which 
view is correct; for this presupposes that we have a test other 
than immediate awareness for deciding which view is correct 
and that our predicament is one of not being in a position to 
apply the test. But our predicament in moral cases is that we 
do not know what other test there is to apply and that we are 
debarred from using the objective-subjective contrast because 
one of the conditions essential to its use is absent. 

The intuitionists are, in fact, trying to combine in one form 
of judgement two assets, objectivity and relative incorrigibility. 
But these cannot be combined since the objective-subjective 
contrast loses its point if objective judgements are made rela- 
tively incorrigible. They use the language of 'really is'; but 
the test that they apply to discover whether something really 
is right or wrong is the test of immediate insight, which is 
analogous to the test for 'looks' and ' feels '.^ 

I. A similar point can be made in connexion with the intuitionists' 
treatment of 'knowledge'. They represent knowledge as incorrigible 
because it is the immediate awareness of a datum which no further 
observation could (logically) upset. But, in fact, a claim to knowledge is 
not of this kind. If I claim to know something as a 'matter of objective 
fact', I shall be amazed or amused or bored if someone proposes to 



If we compare the position of the colour-blind man with that 
of the man whose moral views conflict with received opinions 
the contrast comes out clearly. The colour-blind man is willing 
to admit his own deficiency; but we cannot admit ourselves to 
be morally blind in the same sort of way. The colour-blind 
man will say "It still looks red to me; but it must really be 
brown, since everyone else says that it is brown". But there is 
nothing analogous to this that a moral dissentient can say. 

If I think that a certain type of conduct is wrong when other 
people think it right, there are two things I can do. (a) I can 
say: "I still think it is wrong; everyone else must be mis- 
taken". This is the line taken by the moral reformer and must 
be the line taken by an extreme intuitionist to whom conscience 
is an infallible guide, (b) A less extreme intuitionist might say: 
" I used to think it wrong ; but now that I have discovered that 
many good and wise people think it right I no longer do so; 
I was formerly mistaken". This is the line that would be taken 
by a man who was not confident about his own moral judge- 
ments and was prepared to change them. 

But the one thing I cannot say is (c): "This type of conduct 
still seems to me to be wrong ; but it must be right, since every- 
one says so". And this is precisely what the colour-blind man 
does say in such a case. 

The reason why (c) is impossible cannot be made clear until 
I have given a positive account of what a moral judgement is ; 
but to say that it is impossible may seem surprising. We can, 
of course, admit ourselves to be miserable sinners and also 
admit ourselves to have been mistaken in the past on a moral 
point. We might also admit ourselves to be morally blind in a 
dispositional sense, or recognize a general incapacity to see 
moral truths until someone else points them out to us. We 
may also admit the possibility that our present moral view 
may be mistaken in the sense that we might in the future be 

submit the matter to further tests, since my claim expresses complete 
confidence that such testing can only corroborate my view. But further 
testing is not logically irrelevant. Knowledge claims to be valid; and 
what cannot be tested is neither valid nor invalid. 



led to change it. But what we cannot do is admit that our 
present moral view is in fact mistaken. For to admit this would 
be to abandon the moral view in question. A moral view is, as 
we shall see, a principle that we adopt, an affirmation that we 
are prepared to make ; and to make an affirmation and at the 
same time to admit it to be false is a contradiction-in-use like 
'It's a horse, but I don't believe it is' or 'It's a horse but that 
isn't true'. 

There is one way in which it might appear to be possible to 
admit one's own present lack of moral insight. If a man has 
been strictly brought up in a certain code of morals, he may 
come to reject some article in this code. Thus a Christian, con- 
verted to Islam, might once have believed it wTong to have 
four wives but now no longer does so. But the strength of his 
upbringing might be so great that, even after his conversion, it 
still ' feels ' wrong to him. He has an emotional aversion to the 
practice that cannot be called 'genuine moral disapproval' - 
for he does not disapprove morally any more - but might be 
called 'quasi-moral repugnance'. 

In the same way a reformed teetotaller might still feel veiy 
uncomfortable about having a drink. He might say to himself: 
"^Drinking in moderation is really quite blameless; there is 
really nothing wrong in it, but it still feels wrong to me. My 
moral sense is therefore deficient in this particular ; it does not 
correspond to the real rightness and wrongness of things." He 
contrasts his feeling of repugnance, which is a 'subjective' 
feeling in that he himself is the best judge of whether he feels 
it, with the 'objective' rightness with which it conflicts. And 
this seems to make his case analogous to that of the colour- 
blind man. 

But the appeal to this type of case does nothing to help the 
intuitionist's thesis. For it is not to mere feelings of repugnance 
that he refers when he talks about immediate awareness of or 
insight into objective properties. These are mere emotions, not 
even genuine moral emotions, and they cannot play the part 
which the theory assigns to the apprehension of non-natural 



The Logic of Adjectives 


Fhe study of ethics seems to end in a blind alley. The older 
jhilosophers set out confidently to "erect schemes of virtue 
md of happiness", to discover what the Good Life is or vi^hat 
)ur duties are; but we end with an argument the burden of 
vhich is to show that all their efforts rested on a mistake. In 
jlace of the old, often laborious and sometimes exciting road 
ve are offered the short cut of immediate insight. But our new 
guides not only fail to lead us where we want to go, they do 
lot seem to understand where this is. We ask for help in the 
lolution of practical problems and they offer us a description 
)f a non-natural world. It is not surprising that this has led to 
I radical scepticism in the writings of otherwise very different 
jhilosophers. Both Logical Positivists and Existentialists tend 
deny the possibility of knowledge or rational opinion in 
jthics and to doubt whether we can ever give good reasons for 
ioing this rather than that. 

What has been the cause of this failure? We have been 
logged all along by certain assumptions which belong to logic 
md are implicit in the use of the logician's technical vocabu- 
ary. In this and the following chapter I shall try to show how 
he nature and purpose of practical discourse has been made 
inintelligible by the attempt to elucidate it by means of a 
ogical apparatus unsuited to the purpose and to substitute a 
lew logical apparatus that may be more successful. The first 
rwo assumptions that I shall question are concerned with the 
ogic of single words, especially adjectives; in Chapter 6 I 
shall consider the use of sentences and arguments in practical 

(a) The first assumption is that adjectives are the names of 



properties (or qualities or characteristics), that their logical role 
is that of denoting, referring to, or standing for something.^ 
It is a corollary of this that questions about the meaning of a 
word are to be answered by inspecting the idea, concept, or 
object which it denotes and comparing this with the objects 
denoted by other words. It is as if we discovered the meaning 
of an adjective in the way that a man might discover the mean- 
ings of the names of animals, that is to say by walking round a 
zoo, inspecting the animals in their cages, and reading the label 
on each cage. 

(b) A second assumption, intimately connected with the 
first, is that we can ask what a certain word means instead of 
asking 'What does So-and-so mean by it?'. 

The logician's task is to map the words, phrases, sentences, 
and arguments that we -use in theoretical and practical dis- 
course. And, to perform this task, logicians have elaborated a 
special vocabulary of technical terms such as object, quality, 
property, relation, class, denote, proposition, entail, contra- 
dict, analytic, synthetic, and so on. Like other theoreticians, 
the logician is interested in classifying the objects that he 
studies and discovering what is common to all objects of the 
same type. Text-books of logic usually begin with the assump- 
tion that such statements as 'AH delphiniums are blue', 'AH 
men are mortal', and 'All bishops wear gaiters' are examples of 
a single type which they represent by the formula 'AH S is P' 
And they give to these statements the technical name 'Uni 
versal Affirmative Propositions'. 

Now this procedure of classifying types of words, sentence, 
and argument and giving names of each type presupposes that 
the logician has some method of discovering what words, for 
example, 'stand for properties', what sentences express 'propo- 
sitions' and what strings of sentences constitute 'arguments' 
But logicians seldom have anything to say about this method 
More often they simply assume that all adjectives - (a gram- 
marian's classification) - play the same role and they give to 
this role the technical name 'standing for a property'. 
I. Cf. pp. 31-33, above. 


But we can only understand what this technical phrase means 
if we study the examples that logicians give when they intro- 
duce it. And the examples they give us are very rarely drawn 
from practical discourse ; for they have devoted almost all their 
attention to the logic either of mathematics or of the natural 
sciences. "The various sciences are the best examples of human 
thinking about things, the most careful, clear, and coherent 
that exist. In them therefore the logician can best study the 
laws of men's thinking."^ 

If the logician has done his work well the technical terms 
he uses will elucidate the logic of mathematics or of natural 
science. But it is by no means obvious that the technical 
apparatus he uses will help to elucidate a realm of discourse 
quite different from that from which he chose his examples. 
The making of a decision to do something or of a moral or 
aesthetic judgement often requires careful, clear, and coherent 
thought; but is there any reason to suppose that the logic of 
this thought can fitly be described or criticized by means of 
a technical vocabulary which was not designed to describe and 
criticize it ? Must practical thinking be like the thinking that 
goes on in mathematics or the natural sciences ? 


(a) Do all adjectives stand for properties? Suppose someone 
were to ask whether or not goodness was a property. How 
could this question be answered ? ' Good ' is certainly an adjec- 
tive ; and if to say that goodness is a property is only another 
way of saying that ' good ' is an adjective, it is easy to see that 
goodness is a property. But the doctrine that goodness is a 
property is clearly intended to assert more than this triviality. 
It is intended to assert that the logical behaviour of 'good' 
conforms to the same standard pattern to which the logical 
behaviour of other names of properties conform. But we only 
know what ' standing for a property ' means because we already 
I. H. W. B. Joseph: Introduction to Logic, p. 3. 



know how the typical logician's examples, such as 'blue', 
'loud', and 'round' behave. To say that goodness is a property 
commits us to the very debatable assertion that the logic of 
'good' is like that of 'blue', 'loud', and 'round'. 

The word 'property' is a technical, logician's word and we 
cannot answer the question ' Is goodness a property ? ' in the 
same sort of way that we would answer the question ' Do cats 
eat mice?', by observation. It is more like 'Is a gong a musical 
instrument?', 'Is dancing exercise?', 'Are limericks poetry?', 
'Is medicine a science?'. 

To answer this last question we should have to start with 
certain typical examples that everyone would admit to be 
sciences, chemistry, physics, botany, astronomy, and the like, 
and to compare each in turn with medicine. These standard 
sciences differ from each other in many ways; but they also 
have certain points in common in virtue of which we call them 
all 'sciences'. Since medicine has important points in common 
with the standard sciences and also important differences, a 
straight yes-or-no answer could not be given. Our task would - 
be to discover whether medicine is, on the whole, sufficiently 
like the standard sciences to make it more enlightening than 
misleading to call it a 'science'. And this will depend on the 
associations of the word ' science ' for the questioner. If I tell 
him that medicine is a science, will he expect to discover in it 
Newtonian laws and so come to misunderstand the special 
nature of the doctor's skill ? If, on the other hand, I tell him 
that it is not, will he go away with the impression that medicine 
is 'unscientific' and that medical men are on a par with quacks 
and astrologers? The only safe course seems to be to give 
a longer answer which brings out both the similarities and the 

In the same way we can only hope to answer the question 
'Is goodness a property?' by exhibiting the similarities and 
differences between 'good' and those adjectives that most 
typically fit what the logician has to say about properties ; and 
these are the names of empirical, descriptive properties. 
Moore partially adopts this method. He discovers both simi- 



larities and differences between 'good' and 'yellow'. They are 
alike, he thinlcs, in being simple and indefinable ; but they are 
also so unlilce that we must mark the difference by calling one 
a 'natural', the other a 'non-natural' property. 

Later writers v/ere not content with the vague, negative 
word 'non-natural' and tried to bring out the differences 
between 'good' and the natural properties by calling it a 
'gerundive property'. This was a move in the right direction; 
for the phrase 'gerundive property' begins to bring out the 
intimate connexion between 'good' and overtly gerundive 
words such as 'laudable', 'praiseworthy', 'admirable'. But the 
question they failed to ask was "Why call goodness a property 
at all?". If our previous interpretation of the non-natural 
word theory was correct, Moore intended to mark an impor- 
tant difference in logical status and behaviour between 'good' 
and 'yellow'. Yet this is precisely the sort of difference that 
is denied by calling goodness a property. For what is it to call 
goodness a property but to say that the logic of 'good' is like 
that of other property- words ? The terminology that Moore 
used to mark an important difference that he noticed between 
'good' and other adjectives was singularly ill-adapted to 
bringing out just this difference. 

But, even if goodness is a property, are we justified in 
assuming that the logical role of 'good' is that of standing for 
or denoting that property? " In fact, if it is not the case that 
'good' denotes something simple and indefinable, only two 
alternatives are possible : either it is a complex, a given whole, 
about the correct analysis of which there may be disagreement; 
or else it means nothing at all, and there is no such subject 
as ethics."^ Moore devotes a considerable space to the 
refutation of the first alternative and rightly regards the second 
as unworthy of refutation. But his proof that 'good' denotes 
something simple rests on the assumption that, if 'good' has 
any meaning at all, it must denote something. 

(a) Meaning and Naming. Do all meaningful words and symbols 

I. Principia Ethica, p. 15. 
c 65 


denote? Sharps and flats in music do not denote anything, 
nor indeed does any other bit of musical notation. We might 
toy with the idea that C# 'denotes the fact that' the performer 
is to play G^. Or does it denote the fact that the composer 
had Ciff in mind? But 'denoting the fact that the performer is 
to play C#' is already an odd circumlocution for ' instruction 
to play C#' and the facts denoted are a strange sort of facts. 
Nothing but the assumption that what has meaning must 
denote could lead us into playing such tricks. This assumption 
is obviously f^lse in the case of musical notation. But notes are 
not words and we must see whether there is any more justi- 
fication for it in the case of words. 

To say that a word has a meaning is not to say that it 
denotes something and to say what its meaning is is not to 
say what it denotes. The word 'meaning' itself is both vague 
and ambiguous and depends more than is usual on the context 
and purpose of the speaker. If someone asks what the meaning 
of a word is, he is usually asking for an explanation of the way 
in which it is used. Now the temptation to say that he is asking 
what the word refers to arises' from the fact that the job of a 
great many words is that of referring to something, so that 
to ask how the word is used and to ask what it refers to come 
to the same thing. 

If the word is the name of an ordinary physical object, for 
example 'table', 'mountain', or 'dog', or of an ordinary em- 
pirical quality, 'yellow' or 'round', the easiest way to explain 
its use is to point to the objects concerned or to objects that 
have the quality. But, although this pointing is a way, and a 
very good way, of explaining the meaning of a word, it does 
not follow that what is pointed at is the meaning. 

The typical case of a word of which the job is to refer to 
something is the proper name ; and the typical case of things 
to which we give proper names are people, animals, ships, and 
geographical fe^^tures. These are all ' things ' in the least philo- 
sophical sense of that vague word, visible, tangible, spatio- 
temporal things that you can meet again and again and recognize 
as being the same thing each time. They are also things 



in which we are interested as individuals and not just as 
instances or examples of a general type. Sticks and stones 
could have names too, and would have them if we were inter- 
ested in them as individuals. Some actually do have such 
names ; for example the Rosetta stone, the Rufus stone. 

To equate 'meaning' with 'naming' or 'denoting' is to use 
as a model for all elucidation of meaning the special cases in 
which we should naturally say that a word was the name of or 
denoted something; and the standard case of this is the proper 
name. But the inadequacy of this model is shown by the fact 
that when a word is a proper name we hesitate to talk about its 
'meaning'. Suppose that we were helping a foreigner to trans- 
late a leading article and explained to him the meanings of the 
words 'government', 'statesman', 'constitutional', and so on, 
md that he then asks "What does the word 'Churchill' 
mean?". I think we should reply: "It doesn't exactly mean 
anything; it's a man's name." To say this, with the accent on 
;mean' rather than on 'anything' is not to say that it is 
baeaningless in the way that an accidental blot of ink is 
neaningless. The word has a linguistic job to do ; but its job 
•S one of naming, not of meaning ; and words of which the job 
Is meaning are not names of what they mean. 

(b) Meaning and Context. Elucidating the meaning of a 
vord is explaining how the word is used, and it is only in the 
Jxceptional case in which a word is the name of something, 
;ither a proper name which is used to refer to just one thing 
}r a 'general' or 'common' name that is used to refer in- 
differently to a number of things, that explaining how it is 
ised can be identified with saying or showing what it refers 
o. To treat all words as names of what they mean is to pre- 
uppose that 'meanings' have just those characteristics that 
i thing must have to fit properly into the name - thing situa- 
ion. 'Meanings' come to be regarded as identifiable indi- 
iduals which crop up every now and again, as distinct and 
elf-identical as the boys in a class. And it is because philoso- 
»hers have unconsciously modelled their accounts of meaning 

I ' 67 


on the name - thing situation that they have represented under-' 
standing the meaning of a moral word as an affair of inspecting 
(with the inward eye) the object or quaHty of which it is the 

And this use of the name - thing situation as a model for 
meaning has had a further consequence of the greatest im- 
portance. It has led philosophers to ask the question 'What 
does the word . . , mean ? ' as if it must mean the same thing in 
all contexts. But is it safe to make this assumption ? Ought we 
not rather to ask 'What did So-and-so mean by that word on 
that occasion?' Clearly, if it were never safe to abstract a 
word from its context and to say what it means the writing 
of dictionaries would be impossible. But this assumption is 
only safe in the case of two types of word. 

(i) It is safe if the word is a technical one; for a technical 
language is relatively precise and those who understand and 
use such a language all mean the same thing by the words 
they use. The word ' engine ' does not mean the same thing for 
a schoolboy, an engine-driver, and an engineer, but 'kinetic 
energy', 'erratic block', 'asteroid', and 'differential equation* 
do mean the same thing for the experts who alone use these 
expressions. There is therefore little danger in abstracting 
these words from the context in which they are used and in 
talking about what they mean as opposed to what some par- 
ticular person means by them on a particular occasion. As 
usual, mathematics is at one end of the scale, since mathe- 
maticians insist on a rigour of definition superior to that found 
elsewhere; poetry and other forms of literature are at the 
other end of the scale, where it is most obviously the case that 
we carmot understand what a writer means by a word except 
in the context of his thought as a whole. 

(ii) In the case of common objects, which, as we have seen, 
supply the model which philosophers have used for meaning, 
there is little disagreement about the meanings of words and 
such doubts as arise are mainly doubts about borderline cases. 
But with words that are not the names of common objects 
difficulties of a very different sort arise. For the same word 



can be used on different occasions, not merely as the name 
of slightly different objects, but to do different jobs, some of 
which are not naming jobs at all. With which of these jobs are 
we to identify the meaning? 


As soon as we abandon the assumption that all adjectives 
'stand for properties' we are able to see that there are different 
sorts of adjectives that operate in logically different ways; 
and our problem will be to classify these sorts and to explain 
how they are connected with each other. To say that every 
adjective has its own logic would be to leap from the frying- 
pan into the fire and to confess our inability to make any 
logical generalizations whatever. But this is in fact not the 
case. 'Blue' does not mean the same as 'green'; but there are 
no logical differences in their behaviour and we could classify 
them both as 'colour- words'. Our task will be to examine the 
adjectives used in practical discourse, of which the most im- 
portant are 'right' and 'good', and this must be done with as 
little recourse to the technical terms of traditional logic as 
possible. For a logical apparatus that leads us to say that 
practical discourse is either meaningless or descriptive of a 
special non-natural world must be inadequate. 

For the question 'What does the word . . . mean?' I shall 
therefore substitute the two questions 'For what job is the 
word . . . used ? ' and ' Under what conditions is it proper to 
use this word for that job?' The importance of separating 
these questions v/ill emerge later ;^ for the present I shall 
simply abandon the familiar model of words as labels attached 
to things and treat them as tools with which we do things. 
Talking is not always naming or reporting; it is sometimes 

I. Pp. 90, 95, and 178, below. 



If we examine the adjectives used in ordinary discourse we 
find that they exhibit a great variety of logical behaviour. 
The grammatical form of an adjective sometimes gives us a 
clue to its logical behaviour; for example adjectives in -ent, 
-ible, -ous, and -ic fall into families which differ logically from 
each other, and we can often tell something about the meaning 
of a new adjective from its termination in the same sort of way 
that a chemist could deduce something about a compound un- 
known to him from the fact that its name ended in -ite, -ate, or 
-ide. But termination is not an altogether reliable guide to 
logical behaviour. In the first place words that are not derived 
from Latin or Greek sources seldom have special termina- 
tions, but they do nevertheless fall into logical families; and 
secondly it is notorious that words in -ible and -able function 
in at least two different ways. To say that a man is eHgible for 
parliament is to say that he can (constitutionally) be elected, 
that his election would not be barred by any disqualification 
such as insanity or a peerage ; it is not to say that he ought to 
be elected or is worthy to be elected. On the other hand an 
eligible bachelor is not someone who can legally be married 
but someone worth marrying. 

The classification of adjectives is necessarily a tentative and 
inexact business, especially in a field where it has never been 
attempted by people whose interests are philosophical rather 
than philological ; and for a start I shall distinguish three main 
types. Consider, for example, the following sentences: 

The view from the top was extensive. 

The view from the top was sublime. 
The adjective ' sublime ' does not form part of a description of 
the view, unless we insist on making all adjectives descriptive 
and thereby reduce the force of 'descriptive' to vanishing 
point. We could give an exhaustive description of the view 
by enumerating its contents, and if the list contained a large 
number of large objects this would entail that the view was 



extensive. The question whether the view was extensive or 
not is a question of empirical fact.^ But the subHmity of the 
view is not part of its contents and no description of the view 
would logically entail the truth or falsity of 'the view was 
sublime '. It is just this that the argument against the Natural- 
istic Fallacy shows so clearly. Some philosophers have said 
that goodness is a ' consequential property ', by which is meant 
that it is a property that something can have only if it has 
certain other properties. But we have already seen that the 
link between goodness and the good-making properties is not 
a logical one; a special act of awareness is needed to appre- 
tiend it. The relation between ' sublime ' and those features of 
1 landscape in virtue of which we would call it sublime is of 
:he same type. 
Consider the following conversation : 

A. When I got to the top, I saw the whole plain spread out 
beneath me and Nanga Parbat towering above it. A 
waterfall that must have been at least five hundred feet 
high cascaded down from near where I stood into the 
swirling waters of the Indus. 

B. What a sublime (magnificent, stirring, awe-inspiring, 
wonderful, etc.) sight that must have been. 

B uses the phrase ' must have been ' to indicate that the sight 
vas subUme because of the items in A's description; and A 
vould have been surprised and hurt if B had said: "I don't 
:hink much of that". The connexion between their remarks 
s obviously not logical entailment; yet we feel that B's com- 
nent was the natural and appropriate one to make. And this 
s because he is evincing the natural, appropriate emotion. 
VIost people would react to the description in the same way 
md not say that the view was mean or sordid or squalid. 
Taking a cue from this situation I shall refer to words of 

I. 'Extensive' is not strictly a 'descriptive' word; for to say that a 
aew is extensive is not to describe it. But it belongs to descriptive dis- 
;ourse and will do as an example as I am not interested in the logic of 
lescriptive discourse for its own sake. 



the same family as 'sublime' as Aptness-words (A- words) 
because they are words that indicate that an object has certaii 
properties which are apt to arouse a certain emotion or rang< 
of emotions. I use the word ' indicate ' with deliberate vague- 
ness and do not say, for example, that 'terrifying' could bs"- 
defined in terms of ' causing fear ', Nor, for reasons that wil 
appear later, can we say that 'sublime' or 'terrifying' just ex- 
press the emotion of the speaker. A-words have a logic o 
their own which is different both from that of Descriptive- 1 
words (D-words) and from that of exclamations or reports oj 
one's feelings. 

Consider, again, our use of the word ' weed '. The ordinarj 
man (who is not a Berkeleian philosopher) takes, rightly oi 
wrongly, an uncompromisingly realistic view of D-words. 
such as ' dandelion ' and ' yellow '. He believes that even if there 
were no gardeners there would still be dandelions and thai 
they would still be yellow. But, if there were no gardeners, 
would there still be weeds ? To say that a dandelion is a weed 
is not like saying that it is a member of the order Com- 
positae; and the difference does not lie only in the fact that 
' weed ' is an ordinary-language word. To say that dandelions 
are weeds is not to classify them at all. For the contrast 
between weeds and flowers (in that sense of ' flowers ' in which 
flowers are contrasted with weeds) depends on the interests 
of gardeners. If there were no gardeners we should have no 
use for this contrast; and if the interests of gardeners changed, 
if, for example, dandelions came to be admired for their 
beauty, rarity, or medicinal properties, dandelions would cease 
to be weeds. A weed is, roughly, a plant that we wish to 
eradicate rather than to cultivate. If a man said that he liked 
cultivating groundsel we might think him odd; but if he said 
that he liked cultivating weeds, this would be logically odd 
and we should have to take him to mean that he liked culti- 
vating those plants that others usually wish to eradicate. In 
this way we could remove the logical (but not the horti- 
cultural) oddness from what he says by making 'weed' into 
a descriptive expression. 



For amateur and self-employed gardeners 'weed' is an A- 
word. But for a gardener who is employed by someone else 
its logic is quite diflferent. He may have no interest in his job 
at all or he may like having plantains and dandelions on the 
lawn. For him, a weed is, roughly, any plant that he ought to 
eradicate, that it is his duty to eradicate whether he likes the 
plant or not, and the word 'weed' is a Gerundive-word (G- 
word), roughly analogous to ' praise- worthy ', ' note-v;orthy ', 
'laudable', 'damnable', etc.^ 

The logical relationships between A- words, D -words, and 
G-words wiU be examined later. The present account is over- 
simplified and schematic since it is intended only to throw 
light on the question whether all adjectives ' stand for proper- 
ties '. We might mark the differences by saying that they 
stand for different sorts of properties, aptness-properties, 
descriptive-properties, and gerundive-properties. But what is 
gained by this? 'Red', 'sublime', and 'laudable' are all adjec- 
tives and all obey the same grammatical rules; but we have 
seen that to say that they all stand for properties is to say 
more than this. It is to say that they fit in the same way into 
the same prescribed scheme of categories containing sub- 
stances, properties, states, events, processes, and so on; and 
this in turn is to say that their logical behaviour is similar. 
But this is just what it is not. There is a logical oddness about 
cultivating weeds or being bored by a sublime view that is not 
present in cultivating dandelions or being bored by a view of 
St Paul's. To mark the differences by saying that all adjectives 
stand for properties, but properties of different sorts, is to 
mark it in the wrong way. The intention is to mark logical 
differences, the method is that appropriate to marking logical 

One of Moore's most important arguments seems to depend 
on ignoring the distinction between A-words and D-words. 
"Let us imagine one world exceedingly beautiful. Imagine it 

I . A few more examples : A dress may be red, comfortable, and in- 
decent. A ball may be a leg-break, tempting, and over-pitched. A man 
may be blue-eyed, amusing, and admirable. 



as beautiful as you can ; put into it whatever on this earth you 
most admire. . . . And then imagine the ugliest world you can 
possibly conceive. Imagine it simply one heap oi filth, con- 
taining everything that is most disgusting to us. . . . The only 
thing we are not entitled to imagine is that any human being 
ever has or ever, by any possibility, can, live in either, can 
ever see and enjoy the beauty of the one or hate the foulness of 
the other. "^ The conclusion that Moore thinks we must admit 
is that it would be better for the beautiful world than for the 
ugly world to exist, even though no one ever lived in either. 
But the words I have italicized are not purely descriptive and 
they cannot be understood to mean anything at all if the 
presence of human beings and their tastes and interests are 
excluded, as they must be to make Moore's point. If, for 
example, instead of using the word ' filth ' we specified what 
the second world was to contain in the neutral language of 
chemistry, it is not so obvious that, if there v/ere no one to 
see or smell either world, the one would be better than the 
other. To imagine something as beautiful or ugly, admirable 
or disgusting, is already to ' react ' to it. 

I. Principia Ethica, p. 83. 



The Logic of Sentences and Arguments 


The second pair of traditional assumptions that I shall ques- 
tion is intimately connected with the first pair and is concerned 
with the logic of sentences and arguments. We have seen that 
in the case of the names of common objects and of proper 
names, the meaning of a word can often be given by pointing. 
To understand the meanings of such words it is not necessary 
to consider the parts they play in a sentence or the parts that 
sentences containing them play in arguments. But in order to 
characterize, even roughly, the behaviour of A-words it was 
necessary to illustrate their behaviour by means of sentences 
and even conversations in which they might appear; and this 
will be found even more necessary when we come to give a 
more detailed account of their behaviour. We shall find that, 
although it is convenient to speak of A-words and G-words to 
avoid circumlocutions, it would be nearer the truth to speak of 
A- and G-iises of words or to say that a v/ord can be used with 
an A-force or a G-force. This point is ignored if we make the 
following assumptions: (a) that it is always possible to draw 
a sharp distinction between questions about what a word means 
(or the property it denotes) and questions about the things to 
which the word applies (or the things that have that property) ; 
and (b) that statements can be sharply divided into analytic 
and synthetic. 

These two assumptions together underlie the attack on the 
Naturalistic Fallacy and make necessary the recourse to special 
intuitions. We have already seen how Moore's refutation of 
naturalistic theorists depends on his challenging them to decide 
whether they want to say that ' good ' means ' . . . ' or to say 
that only things which are . . . are good.^ And since he claims 
1. P. 32, above. 



that goodness is unanalysable, he is led to say that propositions 
containing the word ' good ' must all be synthetic. For they can 
never be ' analyses ' of the meaning of the word. 

It is unnecessary to investigate at length the history of the 
distinction between ' analytic ' and ' synthetic ', a history which 
is much older than the words themselves and goes back at least 
as far- as Locke's distinction between trivial and informative 
statements; nor is it necessary to investigate in detail the dis- 
tinctions which logicians rightly draw between analytic, tauto- 
logical, logically necessary, true by definition, and so on. It 
will be enough to say that two concepts or statements are 
said to be connected ' analytically ' if their connexion can be 
established by reference to the logical rules of language and 
requires no reference to the facts of the case. Thus it is analyti- 
cally true that if a man is an uncle he has either a nephew or a 
niece, and ' Fido is an animal ' follows analytically from * Fido is 
a dog'. That John Doe is an uncle or that Fido begs are, on 
the other hand, synthetic truths. We could not establish or 
refute them by appealing to logical rules; we should have to 
examine the facts of the case. 

This sharp distinction between what a word means and what 
it applies to is connected to the assumptions examined in the 
last chapter in the following way. If the role of all nouns and 
adjectives is that of naming, they play only one role. To give 
the meaning of a word is always to say what it is the name of. 
In the case of words that have a complex meaning statements 
that enumerate the items of which its meaning is composed 
will be analytic; but all other statements containing this word 
will be synthetic. 

The need for intuitions arises in two ways. Some of our 
moral knowledge seems to be of a universal kind. Cruelty is 
always wrong; benevolence is always good. This may have to 
be modified when right courses of action conflict or when there 
are two incompatible attributes either of which would be good. 
This may be done by adopting devices such as saying that 
cruelty always tends to be wrong or that there is a prima facie 
duty not to be cruel or to be benevolent. But these are still 



universal propositions. Now it is always agreed that anal)rtic 
or logically necessary propositions must always be true; but, 
if we define ' cruelty ' as a certain sort of behaviour and do not 
use the word ' wrong ' in the definition, the proposition that it 
is wrong must be synthetic. Now moral propositions are in- 
formative and not trivial and they cannot be established by 
reference to the rules of language. How, if this is the case, do 
we know that cruelty alv/ays tends to be wrong ? This is clearly 
not an empirical generalization that we might discover by 
examining a number of cases. It seems that we must have a 
special faculty of insight into necessary, synthetic connexions. 
This is a familiar argument ; but the need for such insight 
also arises in a way that is not so often noticed. When we 
decide to do something or advise someone to do it or make a 
moral appraisal we give reasons. " I ought to pay for his edu- 
cation because he is my son." "You ought to do this because 
you have promised." "He is a bad man because he beats his 
wife." The standard intuitionist account of this situation is that 
the moral properties involved, the obligatoriness and badness 
are 'consequential properties' that we see to be necessarily 
(but synthetically) connected with the natural or descriptive 
properties that we mention in giving the reasons. But consider 
the following conversation: 

A. I should choose that one if I were you. 

B. Why? 

A. Because it's more comfortable. 

The rules for the use of ' should choose ' and ' comfortable ' 
are not such that once we know that one thing is more comfort- 
able than another it follows logically that we should choose it. 
On the contrary we ought often to choose what is less comfort- 
able. The connexion between 'should choose' and 'com- 
fortable' must, so the story goes, be synthetic; and the same is 
true of all the connexions between choosing something and 
the reasons for choosing it. 

Now if this is true it seems that there are only two courses 
open to us, either a complete moral scepticism - there are 



really never any good reasons for choosing this rather than 
that, since the obligation to choose never follows logically 
from any statement about the nature of the thing - or a proli- 
feration of intuitions connecting choosing, advising, and ap- 
praising with the reasons. And there is not so much difference 
as appears at first sight between these alternatives. For what 
is it to claim insight into a necessary synthetic connexion but 
to admit that the connexion is unintelligible ? 
On the intuitionist theory we need : 

(a) an intuition that an action has a right-making charac- 

teristic ; 

(b) an intuition that it is right; 

(c) an intuition connecting (a) and (b); 

(d) an intuition of relative stringencies; 

(e) an intuition that the most stringent obligation is our 


(f) an intuition that we ought to do our duty; 

(g) an inexplicable leap from 'I ought to do X' to the 

executive decision to do it. 

No doubt some of these steps could be represented as analy- 
tic ; but they cannot all be, and it is no wonder that people have 
despaired of being able to give good reasons for doing any- 
thing. Yet it is certainly true that we do give good reasons for 
doing things, if only because the phrase * giving good reasons ' 
is the phrase used in English for something that we all know 
we do ; and if any argument tends to show that this is impos- 
sible there must be something wrong with the argument. 

The need for all these connecting links has arisen because we 
have artificially separated the items to be connected ; and the 
apparent need for this separation arises from the assumption 
that what follows or does not follow from a given proposition 
is the same on all occasions. Naturalistic philosophers have 
often offered definitions of ' good ' and in so doing they have 
not been laying down rules for its use or expressing their in- 
tention of using it in a certain way, but claiming to show how 
it is in fact used. To refute such theories it is only necessary to 



be able to point to one example of a case in which * good ' ob- 
viously does not mean 'pleasant' or 'what I desire', etc., or to 
point out that if a man calls something pleasant it makes sense 
to ask him if it is good, which would not make sense if ' good ' 
and ' pleasant ' meant the same thing. Both these arguments are 
frequently used by intuitionists because they are held to show 
that 'good' and 'pleasant' must be the names of different 

But this argument rests on the assumption that ' good ' always 
means the same thing, which is an assumption that we are only 
tempted to make if we treat it as a name. It shows only that 
'good' sometimes does not mean pleasant, not that it never 
means pleasant. Since the original theory claimed, by implica- 
tion, that ' good ' always meant pleasant, the refutation is fair 
enough. But it scores only a Pyrrhic victory. While it succeeds 
in refuting any simple, one-track theory of what ' good ' means, 
it is powerless to refute any theory which allows that 'good' 
may mean different things on different occasions. Both the 
original theory and the refutation make the same logical mis- 
take of supposing that the meaning of ' good ' can be examined 
in isolation from its context. 


Consider the following conversation : 

A. What are you doing ? 

B. I am having a nice smoke. 
A. Are you enjoying it ? 

The last question is puzzling. Once B has said that he is having 
a nice smoke there seems to be no further room for the ques- 
tion * Are you enjoying it ? '. This is not to say that the question 
is necessarily pointless. Perhaps A has some embarrassing 
communication to make to B and he simply wants to keep the 
conversation going because he knows that it will be more diffi- 
cult to broach the subject if it is allowed to die. Perhaps he 



wants to get B to consider liis answer more carefully. But, in 
the absence of such special reasons, the question seems otiose 
because, in this context, 'nice' expresses the enjoyment of the 
speaker and he has already said that it is nice. 

And, if we suppose that ' nice ' alv/ays functions in this way, 
we might go on to classify ' nice ' as a subjective word and to 
allow that ' this is nice ' is more or less equivalent to ' I like it ' 
or 'I enjoy it'. But this will not do. For there are contexts in 
which 'nice' has little or nothing to do with the enjoyments of 
the speaker. Suppose that someone who had never tasted 
oysters asked if they were nice. A thoughtful person would, I 
think, refuse to give a direct answer to this question. He would 
say "Well, I like them very much, but many people don't, 
especially the first time". If, however, the question had been 
about strawberries most people would answer 'Yes' without 
hesitation. And the reason for the discrepancy is not hard to 
seek. We know that almost everyone likes strawberries and in 
consequence we have good grounds for predicting that the 
questioner will; in the case of oysters we have no such grounds. 
I shall defend the view that the reply ' they are nice ' is, in this 
context, a prediction when I come to discuss A-sentences. For 
the moment I wish only to make the point that ' they are nice ' 
has different implications in different contexts. In the first con- 
text 'nice' was used to express the enjoyment of the speaker; 
in the second to predict the enjoyment of the hearer. This point 
is obscured if we suppose that ' nice ' must mean the same thing 
in all contexts and that what follows analytically from the 
statement ' It is nice ' is the same in all contexts. 

For the concept of logical implication or analytic connexion 
between statements I propose to substitute the concept of ' con- 
textual implication ', and for the concept of self-contradiction 
that of ' logical oddness ' ; and since I shall malce much use of 
these technical phrases they must now be explained. 

I shall say that a statement p contextually implies a state- 
ment q if anyone who knew the normal conventions of the lan- 
guage would be entitled to infer q from p in the context in which 
they occur. Logical impUcations are a sub-class of contextual 



implications, since if p logically implies q, v/e are entitled to 
infer q from p in any context whatever. Contextual implication 
can be most easily illustrated in a case where there is clearly 
no logical implication. 

If Jones says " It is raining", Smith is entitled to infer that 
Jones believes that it is raining, although - Jones believes that 
it is raining ' clearly does not follow logically from ' It is rain- 
ing'. Conversely there is clearly no contradiction between 'It 
is raining ' and ' Jones believes that it isn't raining ', ; yet it would 
be logically odd for Jones to say "It is raining, but I don't 
believe it is". 

The following rules of contextual implication are not rigid 
rules. Unlike the rules of logical implication they can all be 
broken without the speaker being involved in self-contradiction 
or absurdity. 

Rule I. When a speaker uses a sentence to make a statement, 
it is contextually implied that he believes it to be true. And, 
similarly, when he uses it to perform any of the other jobs for 
which sentences are used, it is contextually implied that he is 
using it for one of the jobs that it normally does. 

This rule is often in fact broken. Lying, play-acting, story- 
telling, and irony are all cases in which we break it either 
overtly or covertly. But these are secondary uses, that is to say 
uses to which an expression could not (logically) be put un- 
less it had some primary use. There is no logical limit to the 
possible uses to which an expression may be put; in many 
cases a man makes his point by deliberately using an expres- 
sion in a queer way or even using it in the sense opposite to its 
normal one, as in irony. The distinction between primary and 
secondary uses is important because many of the arguments 
used by philosophers consist in pointing out typical examples 
of the way in which the word ' . . . ' is used. Such arguments 
are always illegitimate if the example employed is an example 
of a secondary use, however common such a use may be. 

Rule 2. A speaker contextually implies that he has what he 
himself believes to be good reasons for his statement. Once 
again, we often break this rule and we have special devices for 


indicating when we are breaking it. Phrases such as ' speaking 
offhand . . . ', 'I don't really know but . . . ', and ' I should be 
inclined to say that . . . ' are used by scrupulous persons to 
warn the hearer that the speaker has not got what seem to him 
good reasons for his statement. But unless one of these guard- 
ing phrases is used we are entitled to believe that the speaker 
believes himself to have good reasons for his statement and 
we soon learn to mistrust people who habitually infringe this 

It is, of course, a mistake to infer from what someone says 
categorically that he has in fact good reasons for what he says. 
If I tell you that the duck-billed platypus is a bird (because I 
' remember ' reading this in a book) I am unreliable ; but I am 
not using language improperly. But if I tell you this without 
using one of the guarding phrases and without having what / 
think good reasons, I am. 

Rule 3. What a speaker says may be assumed to be relevant 
to the interests of his audience. This is the most important of 
the three rules; unfortunately it is also the most frequently 
broken. Bores are more common than liars or careless talkers. 
This rule is particularly obvious in the case of answers to 
questions, since it is assumed that the answer is an answer. 
Not all statements are answers to questions ; information may 
be volunteered. Nevertheless the publication of a text-book 
on trigonometry implies that the author believes that there are 
people who want to learn about trigonometry, and to give advice 
implies a belief that the advice is relevant to the hearer's 

This rule is of the greatest importance for ethics. For the 
major problem of theoretical ethics was that of bridging the 
gap between decisions, ought-sentences, injunctions, and sen- 
tences used to give advice on the one hand and the statements 
of fact that constitute the reasons for these on the other. It was 
in order to bridge these gaps that insight into necessary syn- 
thetic connexions had to be invoked. The third rule of con- 
textual implication may help us to show that there is no gap 
to be bridged because the reason-giving sentence must turn out 



to be practical from the start and not a statement of fact from 
which a practical sentence can somehow be deduced. 

This rule is, therefore, more than a rule of good manners; 
or rather it shows how, in matters of ordinary language, rules 
of good manners shade into logical rules. Unless we assume 
that it is being observed we cannot understand the connexions 
between decisions, advice, and appraisals and the reasons given 
in support of them. 

Logical Oddness. I shall say that a question is ' logically odd ' 
if there appears to be no further room for it in its context 
because it has already been answered. This is not to say that 
the question is necessarily senseless, but that we should be 
puzzled to know what it meant and should have to give it some 
unusual interpretation. In the example of the man having a 
nice smoke it was logically odd to ask if he was enjoying it, 
because in that particular context his previous remark implied 
that he was. 

The task of the moral philosopher is to map the mutual rela- 
tionships of moral words, sentences, and arguments ; and this 
is a task, not of showing how one statement entails or contra- 
dicts another, but of showing that in a certain context it would 
be logically odd to assert one thing and deny another or to 
ask a particular question. 

Consider the following conversations : 

(i) A. I'll have mutton. 
B. Why? 

A. Because I prefer mutton to beef. 

B. Why? (Meaning, not 'Why do you prefer mutton to 
beef?' but 'Why is your preference for mutton a reason 
for choosing it?'.) 

(ii) A. (Picking up sides) I'll have Jones. 
B. Why? 

A. Because he's the best wicket-keeper. 

B. Why? (Meaning, not 'In what way is he the best 
wicket-keeper?', but 'Why is his being the best wicket- 
keeper a reason for choosing him?'.) 



(iii) A. I'll pay the butcher. 
B. Why? 

A. Because I owe him the money. 

B. Why? (Meaning, not 'How did the debt arise?', but 
'Why is the existence of a debt a reason for pavine 

In each of these conversations A starts by expressing a decision 
to do something and when asked 'Why?' he backs up his de- 
cision with a reason. B then tries to insert a logical wedge 
between the reason and the decision. And from the point of 
view of the formal logician he is right. In no case does the 
reason which A gives, which is a statement of fact, entail the 
decision to do what he decides to do. Nevertheless in each case 
B's second question is logically odd. What can he be after? 
What better reason or further reason could he expect to be 
given after the one that has already been given ? 

It should be noticed that, although a reason is given in each 
case, it is a different sort of reason and we shall see that it is 
tied to the decision in a different way. In the first case the 
reason is an expression of subjective preference, in the second 
an objective, verifiable statement of fact, in the third a sentence 
about obligation. The remainder of this book will be largely 
devoted to a discussion of the ways in which these different 
sorts of reasons fit into each other. For the moment I shall 
keep away from specifically moral uses of words and try to 
illustrate the sort of connexions that occur by examining the 
operation of A-sentences. Since the operation of ' good ', ' right ', 
'ought', and G-words is in many respects similar to that o£ 
A-words, the digression is not so irrelevant as it may seem. 



Among the typical A- words are: terrifying, hair-raising, dis- 
appointing, disgusting, beneficial, ridiculous, funny, amusing, 



sublime. There may be a gerundive element in the use of many 
of these words, especially if they are prefixed by 'truly' or 
' really '. Thus, if we call something ' truly disgusting ' we are 
suggesting not only that people are usually disgusted by it, 
but that they ought to be. A man who is not disgusted by some- 
thing that is truly disgusting is not just unusual ; he is in some 
way defective, reprehensible. I shall for the moment ignore the 
G-force of A-sentences. 

It is not to be supposed that all the words in the list operate 
in all respects in the same way. There are logical differences 
between them; but I have grouped them together because they 
exhibit an important likeness. Each of them is connected with 
a specific human ' reaction ', being frightened, amused, cured, 
impressed, and so on. The word ' reaction ' must be widely con- 
strued to cover many things that would not normally be called 
'reactions' at all. It includes attitudes that people might take 
up, emotions they might feel, things that happen to them, and 
things that they do. I shall call the reaction. ' appropriate ' if 
it is the reaction that is logically tied to the A-word in ques- 
tion, for example 'being frightened' to 'terrifying', without 
any suggestion that the reaction ought to be exhibited in a 
particular case. 

To understand the logic of an A-sentence we must ask, not 
'What does it (always) mean?', but 'What does its use in this 
instance contextually imply ? ', ' What would it be logically odd 
to question ? '. The following elements seem to be contextually 
implied in typical uses, although it is not necessary that every 
element should be present on all occasions and the relative 
prominence of the different elements will be different on 
different occasions. 

(a) The subjective element. In default of other evidence the 
use of an A-sentence usually implies that the speaker has the 
appropriate reaction. It would be odd to say that a book was 
enlightening or amusing and then to go on to say that one was 
not enlightened or amused by it. Odd, but not impossible. ' It 
was a terrifying" ordeal, but I wasn't frightened'; 'It may be 
f ery funny, but I am not inclined to laugh ' ; ' It may be a very 



beneficial medicine, but it didn't cure me '. In all these cases 
the subjective element is expressly withdrawn; and, since these 
statements are not self-contradictory, we cannot say that ' X is 
terrifying' either means or entails 'I am frightened by X'. 
Nevertheless, in default of an express withdrawal, we should 
always be entitled to infer that the speaker has the appropriate 

A reaction need not express itself in any overt action ; but it 
must do so in the absence of counter-reasons. For a man to 
maintain that strawberries are nice and never eat them is logi- 
cally odd unless he can give counter-reasons, such as that they 
are bad for his health or against his religious principles. If no 
such reasons can be given he has abused the word ' nice '. 

(b) The predictive element. An A-sentence is sometimes used 
to imply that someone would have the appropriate reaction 
to something if suitable circumstances arose. Thus, if I tell 
you that there is an amusing film at the Scala, when the ques- 
tion is whether you should go to the Scala or not, I am pre- 
dicting that you will be amused. Who the person is must 
depend on the context. In this particular case it would be 
irrelevant to predict the reaction of anyone but you, so that 
I should be breaking Rule 3 if I used ' amusing ' to mean ' apt 
to amuse all or most people ' or ' apt to amuse me '. 

A mountain that is said to be climbable at a meeting of 
experts would not be rightly called climbable at a meeting of 
novices. The statement implies that it would not be beyond 
the powers of those concerned to climb it. 

(c) The generalizing element. Sensible people do not make 
predictions (or retrodictions) except on the basis of evidence, 
so that a man who uses an A-sentence to make a prediction 
contextually implies (by Rule 2) that he has what he believes 
to be good reasons for making the prediction. If he does not, 
he is either lying or abusing language. And what reasons for 
predicting that you will be frightened by X could there be 
other than the fact that people like you are usually frightened 
by Xs? If it could be shown that people were not, in fact, 
frightened by bulls, the statement that bulls are terrifying 



would have to be withdrawn. Terrifyingness is not an ob- 
jective property that bulls would still have even if nobody was 
ever frightened by them, although a particular bull might still 
be terrifying even if nobody had in fact ever been frightened 
by it. 

(d) The causal eletnent. Things are not just terrifying or 
amusing or comfortable; animals are terrifying if they are 
strong, fierce, and malevolent ; plays amusing if they contain a 
high proportion of remarks and situations of a certain sort, and 
cars are comfortable if they have good springs, a heater, and 
plenty of leg-room. To be told that something has an A- 
characteristic is not to be told just what causal properties it has 
that give it that characteristic; but it is contextually implied 
that it has such properties. And if the A-word is sufficiently 
narrow in scope we can often infer what properties it has 
although the speaker does not mention them. 



In what ways are these four elements said to be ' present in ' the 
use of an A-sentence ? Since none of them exhausts the mean- 
ing of an A-sentence and since they do not logically imply each 
other, we are tempted to say that A-words denote special 
causal properties of bulls, jokes, mountains, cars, and so on in 
virtue of which they cause people to have the appropriate re- 
actions. And this temptation is reinforced by the fact that we 
do use A-sentences in an explanatory way. " I laughed, because 
it was funny." "Of course I was frightened; you would have 
been too; it was a very terrifying experience." But although 
A-sentences are used in this explanatory way, the logic of their 
use requires no special properties which they denote; indeed 
such properties only make their use unintelligible. 

The relation between the predictive and the generalizing 



element is clearly inductive and could be expressed without 
the use of an A-sentence at all. 

(i) Most people of type X have been frightened by Ys. 
(3) You, being of type X, will be frightened by Ys. 

This argument is subject to those doubts (real or imaginary) 
which infect all inductive arguments. It is not my purpose to 
discuss the logical gap, if any, between the premise and the 
conclusion, but only to show that it cannot be bridged by the 
introduction of an A-sentence : 

(2) Ys are terrifying to people of type X. 

Such a sentence either does too much or too little, (a) If it 
is held to be both a logical consequence of (i) and to be a 
logical implicant of (3) it converts the inductive argument into 
a deductive one. But clearly your being frightened does not 
follow logically from the fact of others having been frightened 
in the past. 

(b) If both the steps from (i) to (2) and from (2) to (3) are 
held to be inductive, we are worse off than we were before. 
Not only have we now two inductive leaps to make; one of 
them is such that we could never have any reason for making 
it. There are no tests for Ys being terrifying other than the 
fact that people have been terrified by them; hence the fact of 
Ys being terrifying cannot be tied to the fact of people having j 
been frightened by Ys by the observation of a constant corre- 
lation; and such a correlation is of the essence of an inductive 

(c) It might be said that the step from (i) to (2) is quasi- 
analytic, as it surely is. But we are still no better off than 
before. For, on this interpretation, our evidence for the belief j 
that Ys are terrifying is identical with our evidence for the j 
belief that people have been frightened by them. And if we 
are allowed to pass from (2) to (3) we must also be allowed 
to pass direct from (i) to (3); so that the introduction of (2) 

is otiose. 

The temptation to construe A- words as the names of special 


properties does not arise solely from the general tendency to 
treat all nouns and adjectives as nanies. It arises from the fact 
that, while such words obviously mean something, any single- 
track analysis of what they mean is bound to fail. A man who 
says that bulls are terrifying means more than that they frighten 
him. We think of 'being terrifying' as an objective property 
of bulls because 'bulls are terrifying' states a fact which can be 
verified in the way that statements about objective properties 
can be verified. It is true or false independently of the reactions 
of the speaker and it is for this reason that we use the imper- 
sonal form of speech. Moreover, if I am asked why I was so 
frightened, I might reply 'because it was terrifying'; and we 
relax, confident in the belief that the explanation has been 
found. What better reason could there be than the fact that it 
was terrifying ? 

But it is an explanation of a very queer kind. It is not like 
'the bull terrified me because it was strong, fierce, and malev- 
olent ' ; for while this explanation does give the reasons why 
I was frightened, the earlier one does not. It is not the amus- 
ingness of the play that amuses me, but the jokes and situations 
in it. 

Does an A-sentence, then, explain nothing? Is it, like 
* Opium «ends you to sleep because it has a virtus dormitwa\ 
merely a pompous reiteration of the thing to be explained ? It 
does more than this ; but it explains only at a very lowly level. 
The function of an explanation is to remove puzzlement and 
the first step in any explanation is to show that the thing to be 
explained is not really puzzling at all, but only to be expected.^ 
Hence 'They were (I was) terrified on that occasion', is in 
part explained by 'because it was a terrifying situation', since 
the latter implies that most people would have been terrified. 
Their (my) fear was nothing out of the ordinary. 

But A-sentences do more than explain at this lowly level. 

They imply the presence of causal properties without saying 

what these properties are. People do not like Jones because he 

is likeable, but because he is kind, generous, and good- 

I. Toulmin, op. cit., p. 87. 



tempered; and a man who says that Jones is likeable implies 
that Jones has a number of those characteristics for which 
people are usually liked without saying which of these Jones 
has. In the same way people are not able to climb mountains 
because they are climbable, but because they are so shaped as 
to provide sufficient holds for the climber. A-words imply the 
presence of unspecified causal properties because of which a 
thing causes the 'reaction' that it does; but they do not name 
such properties. 

Nor do they name 'consequential properties'. For example, 
being strong is a consequential property of tigers because tigers 
would not be strong unless they had such and such muscles. 
But, in spite of the fact that tigers would not be terrifying un- 
less they were strong, being terrifying is not a consequential 
property of tigers. For 'being strong' is tied to 'having muscles 
of a certain kind' and to 'being terrifying' in different ways. 
The first two are empirical properties between which a corre- 
lation can be observed to exist. But the last is not an observable 
property at all. A man who calls tigers 'terrifying' con- 
textually implies that they have certain properties which 
frighten people. But he does not state this; still less does he 
specify what the properties are. 

It is now possible to see why it was important to distinguish 
the question ' For what job is the word . . . used ? ' from the 
question 'Under what conditions is it proper to use the word 
... for that jobP'.i A-words are used to give explanations 
and to make predictions, not to give the reasons for the 
explanation or prediction. But their use for these purposes is 
only proper if the speaker has reasons of a certain sort which 
are not stated but contextually implied. They are also used 
to express reactions; and it is this fact that tempts us to 
equate 'Xs are nice' with 'I like Xs'. But, even apart from 
the fact that 'Xs are nice' is used for other purposes, its use 
to express a reaction is different from that of 'I like Xs'. The 
use of the impersonal formula would be misleading, and 
therefore improper, if the speaker knew that his taste was 
I. P. 69, above. 


peculiar and that this fact was important. He is not expressing 
a different reaction from that expressed by the personal 
formula; the difference lies in the contextual background. I 
shall try to show later that this distinction sheds light on the 
objective-subjective controversy.^ 

The relation between A-words and the names of causal 
properties is obscured by the fact that we sometimes have no 
special name for the property that causes a certain reaction and 
make the A-word do duty for it. We have, for example, no 
rich, neutral vocabulary with which to refer to the various 
smells and tastes that do, in fact, nauseate, offend, or disgust 
people. Hence we refer to such smells and tastes as ' nauseating ', 
'offensive', or 'disgusting', even when no A-force is intended. 
If we wish expressly to withdraw the A-force we make do 
with a periphrasis that mentions the object that most typi- 
cally has the smell or taste in question : Reeks of disinfectant, 
smells like sulphuretted hydrogen, tastes strongly of paraffin, 
smells of pear-^lrops. On the other hand, since these smells 
and tastes do in fact offend people, the neutral expressions can 
be used with an A- or even a G-force. 'The beer at the Pink 
Elephant tastes of paraffin ' is a sentence that could be used not 
to ' state a fact ' but to issue a warning ; we have no need to add 
that the taste is offensive. But the neutral expressions are also 
used by chemists, doctors, and detectives in a purely neutral 

This brings out the fact that it can be misleading to talk, as 
I have done, of A-words and G-words and, as I shall do, of pro- 
words and con- words. There are no such classes of words; 
there are different purposes to which words can be put and, in 
a special context, almost any word can be put to almost any 
purpose. This is an important feature of ordinary as opposed 
to technical language. But since there are some words that are 
almost always used for a restricted range of purposes I shall 
continue to talk in terms of special classes of words except 
where the context requires a more accurate approach. 
I. P. 178, below. 



Choosing and Advising 


The Purposes of Practical Discourse 


We must now begin to apply the principles outlined in the last 
two chapters and try to understand the role of practical dis- 
course by studying the purposes for which it is used rather than 
by trying to discover (or invent) entities to which the words 
used in it refer. This method has been adopted by philosophers 
in recent years and is, indeed, the natural way out of the im- 
passe into which non-natural qualities led us. To say that some- 
thing is good, we are now told, is not to make a statement 
about it or to describe it, but to express a desire for or an 
attitude towards it, to express approval of it, to grade it, to 
praise it, to commend it, and so on. Philosophers might indeed 
have learnt as much from the Oxford English Dictionary, which 
defines ' good ' not by saying that it is the name of an object or 
property but by its use. It is "the most general adjective of 
commendation" in English. 

We must, however, be careful not to make a mistake ana- 
logous to that of supposing that a given word is the name of 
just one object or property. The commonest practical words 
do not have just one use. They have many uses and can be 
used to do more than one job on any given occasion. But this 
is not the whole story and we must again make use of the dis- 
tinction betv/een the question "For what job is the word , . . 
used?" and the question "Under what conditions is it proper 
to use that word for that job ?". 

If we did not make this distinction we might find ourselves 
compelled to postulate special non-natural uses analogous to 
the non-natural properties postulated by intuitionists. For 
there is clearly a great difference between 'I approve of X' 
and 'X is good'; and if we say that the latter is used to 



express approval we may be compelled to say also that it 
expresses a peculiar, non-natural approval. But the new 
model enables us to bring out this difference without recourse 
to non-natural approval. For we can now say that, while 
both sentences express exactly the same sort of approval, the 
latter is properly used to express this approval only under 
certain fairly stringent conditions that need not obtain when 
*I approve of X' is used. 

W© have already seen that the difference between 'this is 
nice' and 'I like it' (when the former is used to express the 
taste of the speaker) does not lie in the nature of the attitude 
expressed but in the fact that one must not use the impersonal 
formula to express this attitude unless certain conditions ob- 
tain. And the same point applies to the distinction between 
'this is good' and 'I approve of this', a distinction that is all 
too obvious to the objectivist and quite invisible to the subjec- 
tivist. If we distinguish between the purpose for which a sen- 
tence is used and the conditions, if any, which limit the use of 
just this sentence for just this purpose (between what a man 
means by a sentence and what is contextually implied), the 
dispute between subjectivists and objectivists becomes clear. 
Briefly, the subjectivists tend to ignore the contextual back- 
ground and the objectivists tend to treat this background as 
part of what a man actually says. I shall discuss this more fully 
when I come to the word 'good'.^ 

Looking at words as tools with which we do things rather 
than as names or labels which we attach to things has two ad- 
vantages. In the first place it helps to remind us how close the 
connexion is between doing things with words and doing them 
with smiles, nods, winks, and gestures and of the importance, 
in ordinary speech, of emphasis, tone of voice, and manner. All 
these things are more or less irrelevant to the dry, precise, and 
'objective' uses to which mathematical language and the lan- 
guage of empirical science are put, but they are exceedingly 
important in ordinary, and especially in practical discourse. 

I. Pp. 178-80, 232-5, below. 


Ordinary language, well used, is extremely flexible and pre- 
cise ; but the difference between its flexibility and precision and 
that of scientific language comes out in the fact that we never 
use the word ' nuance ' in the latter. 

The second advantage of the new model is that it reminds us 
of the way in which different uses of words shade into each 
other, a point which is obscured by the labelling model since 
things labelled are distinct from each other in a way in which 
the jobs for which a tool is used are not.^ 

Consider the sentences ' What an abominable fellow he is ! ' 
and 'I think he is an abominable fellow'. The first has the 
grammatical form of an expletive, the second that of an indica- 
tive sentence. We are tempted to say that the first expresses 
the speaker's state of mind, while the second states what his 
state of mind is. The first could not be true or false while the 
second could. Yet it is clear that in some contexts the differ- 
ence between them, if any, is very slight, far too slight to 
warrant a sharp logical division. A man who actually said the 
one might, we feel, just as well have said the other. But this 
is not to say that these sentences (always) mean the same 
thing or that they are (always) used in the same way as each 
other. For in other contexts there might be a considerable 
difference, especially if the speaker emphasized the words ' I 

Again: 'What a disgusting thing to do!'. A man might use 
this phrase simply to express his disgust, so that it would 
differ from a curl of the lip only in indicating more clearly what 
he was disgusted at. Yet in other contexts it might mean a lot 
more than this. 'Disgusting' might be at least an A- word, 
contextually implying that his disgust was not peculiar or 

I. The use of mathematics as a model for logic makes the traditional 
terminology of logic singularly inappropriate for elucidating practical 
discourse. In mathematics each expression belongs to a distinct type; it 
must be a numeral or an equation or a function and so on. Mathematical 
uses of expressions do not shade into each other; and mathematics is 
primarily a written, not a spoken language. It uses syriibols, while in 
ordinary language we use not symbols, but words. The inventor of print- 
ing must take the blame for quite a number of philosophical errors. 

D 97 


unusual, or a G-word, implying that his disgust was justifiec 
and that other people ought to be disgusted. 

The fact that a given expression can be used to perforn 
many different roles is by itself sufficient to undermine or a 
least to make us suspicious of all arguments in the form ' Wher 
we say . . . we mean . . . '. And the fact that the same expres- 
sion can do more than one job on one occasion is even mor( 

The old model is not just misleading^; it is wholly wrong 
The words with which moral philosophers have especially tc 
do, which are usually called 'value- words', play many differeni 
parts. They are used to express tastes and preferences, tc 
express decisions and choices, to criticize, grade, and evaluate, 
to advise, admonish, warn, persuade and dissuade, to praise, 
encourage and reprove, to promulgate and draw attention to 
rules; and doubtless for other purposes also. These activities 
form the complex web of moral discourse and our problem is 
to trace the connexions between them and to come to under- 
stand how it is that the same word can be used in all these 
different ways. What a man is doing with a particular value- 
word at a particular time can only be discovered by examining 
what he says in its context, but it would be just as absurd to 
suppose that there is no connexion between these activities 
as to suppose that the same expression can only be used to do 
one job. 

The cormexion between the different uses is partly logical 
and partly a matter of fact; and it is for this reason that the 
analytic-synthetic dichotomy breaks down. It is possible and, 
in some circumstances, not unusual to prefer one thing and 
choose another, to praise a man without approving of him, to 
grade one thing as better than another without feeling any 
desire for, preference for, or interest in either, and so on. Since 
these things are so, we must, if we stick to the analytic- 
synthetic dichotomy, conclude that it is only a 'contingent fact' 
that people desire what they find pleasant, praise the good, 
reprove the wicked, vote for proposals they approve of, etc. 

And so, in a way, it is. We could easily imagine a world in 



which these connexions did not obtain, and this is the sort of 
thing that satirists Hke to do. But it would be a topsy-turvy 
world in which no one would survive for long. Things in our 
world are different; and when a number of logically distinct 
items such as desiring an end, thinking the means appropriate 
to it, and voting for a proposal to adopt those means, normally 
go along together, the 'contingent' connexions between them 
become enshrined in the logic of the language. If a man says 
that a proposal is a good one and holds up his hand with the 
Noes, he is making a logical mistake. 

But this is not because there is an analytic connexion 
between calling something ' good ' and voting for it. He is not 
contradicting himself if he can produce special reasons for his 
apparently inconsistent behaviour. He might, for example, be- 
lieve that the proposal has no chance of passing or that it is 
sure to pass even without his vote and he may have special 
reasons for not coming out openly in its favour. But in default 
of such special reasons, his audience is entitled to infer that 
he would vote for and not against it. 

The same sort of quasi-logical connexion holds between 
approval and exhortation. People do not normally exhort 
others to do or refrain from something unless they themselves 
approve or disapprove of the activity concerned. And, because 
this is so, the same set of words, bad, wicked, vicious, evil, 
etc., can be used both for expressing disapproval and for 
exhorting others to refrain. If a man discoursed for an hour 
on the evils of drinking and used a number of these pejorative 
adjectives, it would be odd for anyone to ask if he really 
disapproved of drinking, unless he were casting doubts on the 
preacher's sincerity. But to do this would be to suspect him 
of using words in a secondary way. It is not just that there 
is a high correlation between the things that a man disapproves 
of, the things he condemns, and the things he exhorts others 
not to do. This correlation is reflected in the logic of our 
language; but not so tightly that we could always convict a 
man of either insincerity or self-contradiction if he said "That 
is wrong, but I don't disapprove of it". 



As Hume pointed out, there are some words which hot] 
describe the way in which a man behaves and also have i 
laudatory or pejorative force. If Jones tells Smith that Browi 
is courageous or honest or kindly, Smith knows that Browi 
can be expected to behave in certain more or less restrictec 
ways in certain circumstances. Jones's remark is therefor( 
partly predictive. But Smith is also entitled to infer that Jone; 
thinks highly of Brown. It is because it is usual to commem 
and not disparage the types of behaviour described by these 
words that there is an air of self-contradiction about calling 
a man brave, honest, and kindly and refusing to praise him 
for these words are terms of praise. 

Since I shall make frequent use of the principle that a giver 
word can not only do two or more jobs at once but also h 
often, in the absence of counter-evidence or express with- 
drawal, presumed to be doing two or more jobs at once, I shal 
give it the special name ' Janus-principle '. 


Among the various performances for which value-words arc 
used, which shall be given pride of place ? To some extent this 
is, as in an axiomatic system, a matter of choice. But since ] 
started by posing the fundamental question as 'What shall 
I do ? ', it will be convenient to start with choosing, deciding, 
and preferring. 

Choosing is something that we do. It can be done without 
using any words at all, as when a man chooses a card to play 
or an apple from a plate of fruit. It can be done with words, 
as when a man in a shop says " I'll have that one ". This is not 
a prediction; for it would be odd if the shopkeeper repHed 
" I don't expect you v/ill ", though he might without making 
a logical mistake reply: "No, you won't" or "That one is not 
for sale". Such formulae are not statements but expressions 
of choice; they are as much interventions in the world as a 
non-linguistic action is, and they differ from non-linguistic 

100 ^ 


actions only in that they produce their effects in a conventional 

Choosing, deciding, and doing. In some cases we choose with- 
out doing anything. For example a man can choose a place 
for his holiday next year. In this case we may equally say that 
he has decided or made up his mind where to go. The decision 
is, as it were, put into cold storage to be taken out and acted 
on when the appropriate time comes. Although it is quite 
proper to use ' choose ' in this way, I shall limit my use of it to 
cases where choosing involves doing something at the time. 

Since choosing involves doing something, a man cannot be 
said to have chosen until he has at least started on his chosen 
course. Merely to decide to play the ace of spades is not yet 
to choose it, and a decision does not therefore entail a choice. 
But the relation between the two is, nevertheless, not a con- 
tingent one. The fact that people choose to do what they have 
decided to do is not something that we discover by observa- 
tion of human behaviour. The relation is one of quasi-impli- 
cation. A man may decide, for example, to vote for Jones and 
be prevented from doing so by sickness or the cancellation of 
the election. But, in default of such explanations, there is no 
logical gap between deciding and doing. 

If a man decides to do something in the future he may fail 
to do it because he has changed his mind; but it would be 
absurd to say in this case that he had not really decided. 
Decisions are none the less genuine for being revocable. A 
change of mind may be sudden and inexplicable. In this case 
ex hypothesi no reason can be given for it, and if a man decides 
to do X and suddenly changes his mind and does Y he acts 
irrationally. (There are other ways in which actions can be 
called ' irrational ', for example if the agent has no good reasons 
for believing that they will produce the end which he desires ; 
and in most uses ' irrational ' is an opprobrious word. I do not 
wish to imply here that every action which deviates from a 
decision in an inexplicable way is irrational in any of the other 
senses or to imply that we ought never to act in this way.) 

A change of mind can, on the other hand, be rational; the 



old decision is scrapped and a new one made on the basis of 
new evidence. But 'choosing' and ' deciding ' are used in such 
a way that it is analytic to say that if a man has decided to do 
something and does not do it then either he was prevented or 
he changed his mind. If neither of these explanations of the 
disparity between the decision and the choice can be given, the 
man who in fact did X cannot be said to have decided to do Y. 

Choosing and preferring. A man can choose without having 
any reason for his choice in sheer absence of mind or from 
sheer force of habit. These are minimal cases that hardly 
deserve the name of ' choice ', and I shall consider only cases 
where a man has reasons for his choice. 

The word 'good' can be used to express a preference and 
when so used is always a concealed comparative. Part of its 
function is to compare the thing called ' good ' with others not 
so good, although the comparative element need not be promi- 
nent and no explicit comparison may be in mind. If chal- 
lenged, however, the speaker must be prepared to make such 
a comparison. When used in the context of choice the com- 
parative element is prominent, since choice is always between 
two or more alternatives. Preference begins with thinking 
this course better than that and ends with deciding that it is 
the best, and the sentences that we use to express preferences 
are tied to those used for expressing decisions in the same 
quasi-analytic way that the latter are tied to doing. There is 
no need therefore to try to bridge the gap between ' this is the 
best thing for me to do ' and ' I shall do this '. In deciding that 
something is the best thing for him to do a man has already 
decided to do it. 

The first sentence expresses a preference for one course of 
action over all others, the second a decision to take it. Choice 
would be an unfathomable mystery if these two sentences were 
interpreted as being only 'synthetically' connected with each 
other. A mysterious gap would always emerge; for it would 
make sense to say: "I prefer this course to any other (or this 
is the best course for me), but what shall I do?". These 
collocations of sentences are logically odd, but not self contra- 



dictory since, as we shall see, they could have a point in 
some circumstances. And recognizing the logical oddness of 
the question does not consist in having insight into a neces- 
sary connexion between preference and choice but in being 
puzzled to know what the question could possibly mean. In 
saying that he prefers A to B a man shows that he has already 
solved his problem of choice: yet in asking the question he 
appears still to have a problem. He can, of course, continue 
to ask himself " What shall I do ? " even after he has decided 
that X is the best thing for him to do. But if he does so he is 
using words parrotwise or pretending to have a problem to 
solve when in fact he has not. The question cannot bear its 
usual sense; and this is a logical 'cannot'. 

When I deal more fully with 'good', I shall examine the 
view that a man chooses to do something because he thinks it 
the best thing for him to do and try to show that 'thinking 
something is the best thing to do ' is not a reason for doing it, 
because it is too good a reason.^ We must now consider some 
cases in which it would not be logically odd for a man to prefer 
one thing and choose another. 

(i) A man may prefer travelling by road to travelling by 
rail but in fact always choose to go by rail. But he can (logic- 
ally) only do this if he is able to offer some explanation. For 
example he might be an employee of the railway company and 
get a free ticket, or his mother might think road travel very 
unsafe and he wants to humour her. As long as he can explain 
the discrepancy in some such way there is nothing logically 
odd about his saying that he prefers one thing when he always 
chooses the other. His preference entails only that he would 
travel by road if other things were equal. But in default of 
such explanations we should be entitled to infer that he did 
not really prefer going by road, whatever he may say. Our 
choices are better evidence of our preferences than our words. 

(2) A man may prefer peaches to apples in general but 
choose an apple on a particular occasion, if, for example, there 
are not enough peaches to go round. What he prefers in this 
I. P. 161 below. 


case is not the apple to the peach but the total situation in 
which he has an apple and someone else has a peach to that in 
which he has a peach and someone else has an apple. But again, 
a man could not be said to prefer peaches to apples if he 
regularly chose apples without being able to advance any 
explanation. It makes sense to say: "I know you prefer 
peaches; why did you choose the apple?"; but it is logically 
odd to say: "I know you prefer peaches; but why did you 
choose the peach?". 



Reasons for Choosing 


In this and the following chapters we shall be coRGerned with 
the vocabulary used in deliberating, for expressing decisions, 
and for giving reasons for our decisions. The phrase 'good 
reason' is ambiguous. It may mean 'morally good reason', 
that is to say a reason which justifies an action and exempts 
the agent from moral censure ; or it may mean ' logically good 
reason ', that is to say a reason which leaves no further room 
either for the question ' What shall I do ? ' or for the question 
'Why did you (he) do that?'. I shall use the phrase 'good 
reason' in the latter sense. In accordance with the principles 
laid down in the preceding chapters our task is not to discover 
propositions that entail a decision to act but propositions which 
are such that, once they are granted, it would be logically odd 
either to ask for further reasons for doing something or for 
a further explanation of why someone did it. 

Precisely the same vocabulary is used both in deliberating 
ante rem and in explaining post rem why someone did what he 
did. If 'because . . . ' is a logically good reason for doing some- 
thing it also constitutes a logically good explanation of why 
someone chose to do it. 

The questions involved are logical, not psychological, still 
less physiological, ones. We are not at present concerned with 
questions about the general types of thing that people choose 
or choose to do. For example. Do people prefer oranges to 
lemons ? Is fear a more powerful motive than greed ? Still less 
are we concerned with physiologists' explanations correlating, 
for example, excess of adrenalin with anger or a craving for 
sugar with certain defects in metabolism. This point is impor- 
tant since philosophers have often written as if they were 



enunciating empirical laws of human behaviour, which they 
are not usually competent to do. They have said, for example, 
that men only desire pleasure, and they have put this in a way 
that leads one to suppose that, in their belief, men might 
(logically) desire all sorts of things but in fact never desire 
anything but pleasure. Since we all believe that we often desire 
food and drink and other things, the philosopher appears to 
be convicting the whole human race of a colossal mistake about 
its motives; he appears as a sort of Super- Freud, telling us 
that our motives are always quite other than what we thought 
they were. 

But it is clear from their arguments that philosophers are 
really doing no such thing. What look like generalizations 
turn out to be logical observations ; and these are either recom- 
mendations to adopt a certain set of concepts with certain 
logical relations between them or assertions to the effect that 
such a system already underlies, though unnoticed, our 
ordinary use of these concepts. In the latter case what the 
philosopher says can be verified or falsified, but it would be 
by an examination of linguistic usage, not by an examination 
of non-linguistic behaviour. 

Are all the reasons that can be given for choosing to do 
something of the same logical type ? Do they all enter into thie 
logic of deliberation and explanation in the same way ? Or do 
they play different roles, comparable to the different roles 
played by laws and facts in giving an explanation of a natural 
phenomenon ? Perhaps the most celebrated attempt to explain 
all reasons for doing or refraining from something on a single 
model is that of Hobbes, according to whom all action is caused 
by desire or aversion.^ It is worth while examining this 
theory at some length. I shall try to show later that the usual 
criticisms of it are misplaced; it is, however, open to a fatal 

The word 'desire' is commonly used in connexion with 
food, drink, and sex. We also talk of a desire to scratch, to 

I. Leviathan, chapter 6. A desire is an 'endeavour toward' and an 
aversion is an 'endeavour fromward'. 

1 06 


smoke, or to be rid of some uncomfortable, irritating, or pain- 
ful sensation. Any of these desires, when very strong, would 
be called a 'craving', and this will turn out to be important, 
because it helps to explain our talk about 'being a slave to 
one's desires ' and so explain why some philosophers have held 
the paradoxical theory that to act from desire is never to act 

One necessary element in the situations mentioned is that 
(a) there must be a bodily sensation or feeling of some sort. 
It is also necessary that the sensation should be (b) painful, 
distressing, uncomfortable, irritating, or unpleasant, and (c) 
that we should want to be rid of this sensation. And it is 
because we want to be rid of it that we take steps to remove 
it. This pattern of behaviour is so familiar that it needs no 
further description. If anyone says that he does not know what 
a ' desire ' is, in this limited sense of the word, there is nothing 
that can be done for him. He knows well enough - I shall 
assume - what the facts are. The difficulties arise over the logic 
of the concepts used in discussing this type of situation. 

The logical connexions between the unpleasant (etc.) sen- 
sation, the desire to be rid of it, and the taking of steps to be 
rid of it are so intimate that they make certain questions 
logically absurd. To ask someone who is in pain why he wants 
to alleviate his pain and to ask why someone who wants to 
alleviate his pain takes steps to do so are to ask logically odd 
questions; there is no room for them, unless they can be 
construed in an unusual way. But this is not to say that the 
connexions are analytic, that we could define, for example, a 
' painful ' sensation as one that a man wants to get rid of or 
'wanting to get rid of as a tendency to try to get rid of. The 
logical difficulties involved are, as usual, due to the fact that 
certain empirical connexions are so common and widespread 
that they have come to be enshrined as logical connexions in 
our talk. Almost all the v/ords that I have used in this pre- 
liminary sketch are Janus-words. 

We have no neutral language for talking about those sen- 
sations that, as a matter of fact, most people find painful. If 



asked to describe my sensation I must either use a periphrasis, 
such as 'as if someone was sticking pins into me' or *as if 
someone were hitting me on the head with a hammer' or I 
must use a word such as * ache ', ' throb ', ' tremor ', ' palpitation ', 
all of which, simply because most people find these painful on 
most occasions, already carry an overtone of painfulness. 

These words can be used in a purely descriptive way to 
convey information about the sort of sensation that I am 
having; and, when so used, they carry no implication about 
my wanting to get rid of the sensation. Logically I could enjoy 
an ache or an itch and want it to go on ; but the fact that most 
people usually do not tends to make us suppose that when 
someone says that he is itching or that a gnat bite irritates 
him he must want to remove the irritation. And this con- 
nexion then becomes a quasi-logical one, because the desire 
to get rid of the itch is contextually implied. 

And this is especially true of the word ' pain ' and its deriva- 
tives. If a man says that a sensation is painful we are entitled 
to infer that he wants to get rid of it. He may, of course, put 
up with it for the sake of something else. But as far as it goes 
the assertion that something is painful, irritating, unpleasant, 
or a source of discomfort implies that the speaker would rather 
be without it. 

The phrase ' source of discomfort ' is particularly dangerous ; 
for it resembles ' cause of irritation ' and thus tends to make 
us think of ' discomfort ' as itself a kind of sensation. And this, 
as we shall see in the next chapter, is a fatal error. When a 
man ' feels discomfort ' or ' is uncomfortable ' he does not have 
two sensations, an itch (or whatever it might be) and a special 
sensation of discomfort ; he has only one sensation which is an 
uncomfortable one. Similarly, when a man experiences a pain- 
ful sensation he has one sensation, not two. The painful sen- 
sation of having a pin stuck into you cannot be analysed into 
two sensations, a neutral pin-going-in sensation and a sensa- 
tion of pain. Indeed the philosophical phrases 'sensation of 
pain ' and ' sensation of pleasure ' are fatal corruptions of the 
ordinary phrases ' painful sensation ' and ' pleasant sensation '. 



And it is still more fatal to represent desires as sensations. 
The desires I am considering all involve sensations in so 
intimate a way that it is natural to try to identify the desire 
with the sensation involved. Thus we can say that hunger is 
' a desire to eat ' ; but hunger is also the sensation that you feel 
in the pit of your stomach. So it follows, then, that a desire 
to eat is a sensation in the pit of the stomach ? Not a bit of it. 
As is always the case with a complex of different sorts of 
elements, the same word or phrase can be used to refer to the 
whole complex situation or to any important element in it. 
* I feel hungry ' is a Janus-phrase that both refers to my sensa- 
tions and also expresses a desire to eat. And it is used with 
this double force because it is an empirical fact that people 
who have the sensation referred to also want to eat. 

Is fear a complex of sensations, for example trepidations, 
nausea, and a quickening pulse ? Or is it a desire to run away ? 
Or is it a tendency to run away? These are idle questions; 
for ' fear ' can be used to refer to any of these things, and the 
logical fact that it is a Janus- word is due to the empirical fact 
that when people feel these sensations they usually v/ant to 
run away and that, ceteris paribus, when they want to and can 
run away they usually do. So, if a man stands his ground with 
no counter-reasons for doing so in a situation that would make 
most people afraid, we say that he is not afraid ; and this seems 
to entail that he did not have the sensations to which 'being 
afraid' refers. But in fact, of course, there is no such entail- 
ment. He may be having the sensations and be enjoying them. 
Now the fact that he enjoys the sensations concerned may be 
expressed either by saying that he enjoys these painful sen- 
sations (thus tying ' painful ' logically to the sensations) or by 
saying that he does not find the sensations painful (thus tying 
'painful' logically to non-enjoyment). And whichever way of 
putting it we choose we are inclined to suppose that a logical 
puzzle arises because what we say appears to be self-con- 
tradictory. But this is, of course, not so. We know that 
masochism occurs. 

But we also talk of ' desires ' in cases where there is not, at 



any rate not obviously, any painful, distressing, or unpleasant 
sensation that a man wants to get rid of. Hobbes' definition of 
'Ambition', for example, as "desire of office or precedence" 
does not sound strained or unnatural. But in this case there 
is no sensation that the ambitious man wants to get rid of, 
at least so long as he is successful. 

We certainly speak of ambition as 'gnawing' and this 
metaphor, which assimilates ambition to the itch-scratch 
pattern, must be explained. Thwarted ambition is commonly 
accompanied by feelings (not necessarily bodily sensations) of 
anger, regret, mortification, and spite. Indeed it must be so 
accompanied, since we should refuse to call a man 'really 
ambitious ' if he had none of these painful feelings when he 
failed to achieve his aim. But although feelings are connected 
with ambition, they cannot be connected with it in the way 
that sensations are connected with hunger ; for they arise only 
when the ambition is thwarted. Hunger might not unnaturally 
be called a desire to remove certain unpleasant sensations and 
might also be called a desire to eat ; but no such double account 
could be given of ambition, since the feelings involved only 
arise when ambition is thwarted and ambition cannot there- 
fore be construed as a desire to eliminate antecedent unpleasant 

And there are many cases of wanting to do something from 
which antecedent sensations are even more conspicuously 
absent. A man who wants to smell a rose or to listen to a 
symphony is not disturbed by pains or discomforts of any 
kind, though, as in the case of the ambitious man, he may feel 
annoyance or regret if he fails to get what he wants. More- 
over we have a number of expressions which are like ' desiring ' 
and 'wanting' in that they provide logically impeccable ex- 
planations of choice but unlike them in that they do not 
obviously involve a reference to any feelings at all. These are 
'being fond of, 'being interested in', 'having a passion for' 
(which, of course, does not involve being in a passion about 
anything), 'taking pleasure in', and 'approving of. 

These expressions are all used in such a way that it would 


be logically odd to go on asking why a man was doing some- 
thing once it was established that he was interested in it or 
fond of it, etc. This is the point they have in common with 
' desiring '. But it would be a mistake to conclude from the fact 
that they explain an action in the same sort of way that a 
reference to desire explains an action that they are all desires. 
For to do this is to suppose that all these explanations must 
conform to the same pattern, and this tempts us to invent 
logical analogues in these cases for all the elements that must 
occur in cases of desiring. And it is just this mistake that 
Hobbes makes. By calling all motives 'desires' he invites us 
to believe that motive-explanations all conform to one pattern 
and specifically to the itch-scratch pattern; but to do this is 
to distort the logic of many of these explanations and to saddle 
our account of motives with fictitious elements. 


Hobbes' treatment of all motives as desires is open to criti- 
cism; but it cannot be condemned merely on the ground that 
it conflicts with the ordinary English use of the word ' desire '. 
Ordinary language, being used by ordinary people on ordinary 
occasions, is not rich in highly abstract or generic nouns; we 
have, in fact, no very general word to cover desiring, liking, 
wanting, approving, enjoying, being fond of, interested in, 
and so on ; and part of the reason for this is that these concepts 
are in many ways dissimilar. But it does not follow that 
Hobbes' general distinction between ' endeavour toward ' and 
' endeavour fromward ' is altogether a mistake ; and if there is, 
as I shall try to show, a good reason for making this general 
distinction, our only course is either to invent a new generic 
word or to adopt the more risky course that Hobbes adopts 
of selecting a word which normally has a narrow use and 
using it in a novel and wider way. 

The only ordinary English word that is sufiiciently vague 
to be made to cover (with a little terminological ingenuity) 



all the cases considered is 'wanting'; but this word, just by 
reason of its vagueness, is far too weak to cover the more 
violent passions or the most permanent and deep-seated 
desires. (It v>^ould be absurd to talk of a Christian 'wanting' 
to inherit the kingdom of heaven or of a near-hysterical 
bobby-soxer ' wanting ' to see a crooner.) 

With some hesitation I have chosen the expressions 'pro-' 
and ' con-attitude ' to fulfil the required role ; but it must first 
be shown that there is a need for any generic term whatsoever. 
Consider the following lists: 

List A List B 

Like Hate 

Approve of Dislike 

Enjoy Disapprove of 

Love Detest 

Want Reject 

Accept Decline 

Pleasure Shy away from 

Comfort Pain 

Desire Discomfort 

Happiness Aversion 

Interested in Try to avoid 

Fond of . Try to prevent 

Try to achieve Try to get rid of 

Try to acquire Try to stop 

Try to obtain Try to avert 
Try to prolong 

I have deliberately included in these lists concepts of different 
logical types in order to make clear that the use of the phrase 
'having a pro-attitude towards' should not be restricted to 
any one pattern ; and I am, for the same reason, unmoved by 
the criticism that many of the items on the list would not 
normally be said to have anything to do with ' attitudes ' at all. 
The word ' attitude ' has been selected just because it is vague. 
The important point that I wish to bring out lies in the words 
'Pro' and 'Con'. 

In spite of the heterogeneity of the lists, it can be shown 



that there is a genuine basis for the distinction by remarking 
that the reader would have no difficuhy in most cases in saying 
into which hst, if either, some new concept ought to be put. 
I have also restricted the lists almost entirely to verbs; for, 
in most cases, to include nouns or adjectives would be to 
prejudice both logical and empirical issues. I might, for 
example, have included ' good ' and ' bad ' in the lists, for their 
places are obvious. But this would prejudge the issue whether 
to think something 'good' just is to have a pro-attitude to- 
wards it, in which case 'thinking good' should clearly be 
included, or whether thinking something good is not rather to 
suppose it to have some quality which is more remotely 
connected with having a pro-attitude. 

To include nouns such as ' health ' and ' disease ', ' riches ' and 
' poverty ', ' peace ' and ' war ' would be to prejudge an empirical 
issue and to suppose that men always have a pro-attitude to 
the first of these pairs and a con-attitude to the second. 



Sentences containing pro- and con-words provide good - that 
is to say, logically complete - explanations of choice. If you 
ask a man why he is gardening or why he is going to turn 
.on the wireless and he says that he enjoys gardening or wants 
to hear some music, he has given a reply that makes a repeti- 
tion of the question logically odd. It must always be remem- 
bered that a repetition cannot be ruled out as senseless; the 
situation is that it could not have its ordinary meaning and 
would have to be interpreted in a special way which will 
depend on the context. For example, by repeating the question, 
the questioner might be trying to insinuate in a delicate way 
that the reason given was not the true one or that it was a 
morally reprehensible one. But in default of such special 
interpretations it seems senseless to ask anyone why he is 
doing something when he has told you he enjoys it, likes doing 
it, or wants to do it. 



Pro-words differ from each other in many ways, but they 
all have this in common : that they provide logically impeccable 
explanations of why someone chose to do something. They 
also provide logically impeccable reasons for deciding to do 
or not to do something. The 'reason for doing' which is ex- 
pressed by such a phrase as ' because I want . . . ' or ' because 
I enjoy . . . ' may be counteracted by other and more weighty 
reasons for making the opposite choice ; but each pro-sentence 
refers to a reason which, in the absence of counter-reasons, it 
would be logically odd not to choose. 

Many philosophers have made the point that every action 
must have a motive, and that a motive can only be counter- 
acted by another motive; and some have represented choice 
as simply the victory of the strongest motive or set of con- 
current motives. These points have usually been put as if they 
were psychological laws; but they are really elucidations of 
the logic of concepts. To say, for example, that every action 
must have a motive is to state a tautology, since what a man 
'did' without a motive would not count as an 'action'. The 
theory that a motive can only be counteracted by another 
motive is also a logical rather than a psychological theory. For 
it is the theory that we use the word ' motive ' in such a way 
that anything which counteracts a motive is also called a 
motive. In the same way, my lists of pro- and con-attitudes 
must be so construed that anything which could be offered as 
a logically good reason for or against doing anything must be 
included in the lists. By a ' logically good reason ' I do not mean 
a morally good reason; I mean anything which, when offered 
as an explanation of why someone chose to act as he did, has 
the force of making further questioning logically odd. 

The proposition that any statement which gives a logically 
complete reason for choice must include a reference to a pro- 
or a con-attitude is thus a frank tautology. Its function is to 
draw attention to the existence of a class of words that have 
just this force. My purpose is not to prove an avowed tauto- 
logy, but to show that many expressions which are nor- 
mally used in giving reasons for choice do not give logically 

114 A 


complete reasons. They are given and taken as complete 
explanations only because certain assumptions are involved 
when they are used. In particular I shall try to show that 
deontological words (right, obHgation, duty, and their cog- 
nates) never give logically complete reasons for choice. To 
put the point in a paradox which I shall defend later, ' believing 
that something is the right thing to do' and 'believing that 
something is my duty' are never good reasons for doing it 
and such beliefs never explain why people do what they do. 

It is important to notice that a reference to a pro- or con- 
attitude provides a good (logically complete) reason even in 
cases of morally reprehensible conduct. If a man says that he 
is fond of music he has fully explained why he listens to music 
in a way that makes it senseless to say " I know you are fond 
of music; but why do you listen to it?". And this is equally 
true if he says "because I enjoy it" when someone asks him 
why he is torturing the cat. The reply does not give a morally 
good reason ; but logically it is impeccable. 

To ask why people enjoy those things that they do 
enjoy is to ask a psychologist's question which could be an- 
swered. (Hobbes, for example, propounds the psychological 
theory that we desire to do those things that help or fortify the 
"vital motions".) To ask whether people ought to enjoy what 
they enjoy is to ask a moral question; and this also can be 
answered. But to ask why people do the things they like and 
enjoy doing is not to ask a question but to raise a gratuitous 
philosopher's puzzle. 

It is important to notice that there are no logical limits to 
the possible objects of pro- or con-attitudes, other than the 
logical limits of language itself. This is important since philo- 
sophers who have made pro-attitudes such as desire or 
approval the central concepts of their ethical system have 
often been accused of holding special (and very unplausible) 
theories about the sorts of things that people can desire or 
approve of. I shall try to show in later chapters that the 
tendency of some modern philosophers to accord a peculiar 
and prominent position among motives to the sense of duty 



is due to this mistake. The sense of duty is indeed peculiar 
and important; but its peculiarity and importance are not what 
they have been thought to be ; and the current view of it leads 
to paradoxes and absurdities. But this interpretation can be 
avoided if we remember that to explain actions in terms of 
pro-attitudes entails no view whatsoever about the objects of 
these attitudes. 


We must now consider certain apparent exceptions to the rule 
that every complete explanation of conduct must refer to a 
pro- or con-attitude. 

I. Means and Ends. If a man does something, not for its 
own sake, but because he knows that unless he does it he will 
not be able to do something that he has a pro-attitude towards 
doing, it is not necessary that he should have a direct pro- 
attitude towards doing it as such. A man who buys concert 
tickets need neither like nor enjoy nor approve of buying 
concert tickets. He explains his action, not by saying that 
buying a ticket was just what he wanted to do, but by saying 
that without a ticket he could not go to the concert. But, in 
this case, it is clear that his action is not fully explained until 
he has stated the end towards which it is a means. It is only 
when he says " I like music " or " I wanted to hear the Emperor 
Concerto" that he has fully explained his action in a way 
which makes it senseless to say " But I still don't understand 
why you bought the tickets". 

There is some point to the dictum that a man who wills 
the end must will the means. This is absurd if it is taken to 
mean that a man who wants something necessarily also wants 
to get it by hook or by crook. If, for example, the only way 
of getting tickets is to steal them, he may still want to go to 
the concert, but his choice is between getting-tickets-by-stealing 
and not-getting-tickets-by-stealing, and he may well choose 
the latter. But unless he has a con-attitude towards getting- 



X-by-doing-Y, it is irrational for him not to do Y. This use 
of 'irrational' is another example of the way in which the 
logic of language reflects obvious empirical facts. It is not, in 
Hume's sense, " contrary to reason " for him to refrain from 
Y; but it is silly. And the logic of the language we use in 
talking about human action is built up on the assumption that 
men choose to do what they have a reason for doing and 
choose means which they believe to be effective. This is why 
we use the logical words 'reasonable' and 'unreasonable' to 
characterize actions. 

2. Statements of fact. In ordinary life it is often sufficient 
to point out the facts of the case or to single out some special 
fact in order to explain someone's action. Thus a proper 
answer to 'Why did Jones help that man across the road?' 
would be ' Because he is bhnd'. But such an answer is no more 
complete than 'because he is a philosopher' is a complete 
answer to 'Why is Socrates so inquisitive?' or 'because he's 
the Mayor' a complete answer to ' Why is that man wearing 
a chain ? ' In these cases the answers take the form of pointing 
out a fact which explains the thing concerned only via the 
generalizations 'All philosophers are inquisitive' and 'AH 
mayors wear chains'. The very fact that the particular pro- 
position is given as an explanation contextually implies that 
there is such a general rule; for only so could the particular 
fact alleged be an explanation. Equally we could have an- 
swered 'because all philosophers are' or 'because all mayors 
wear chains', and had we done so our offering a general rule 
in explanation would have entitled the hearer to conclude that 
Socrates is a philosopher or that the man was a mayor. In each 
case the explanation requires both a general rule and a par- 
ticular proposition, and it is because this form of explanation 
is so common that we do not need to mention both. Which 
we choose to mention is determined in each case by our 
opinion as to which point is most likely to be unknown to the 

In the same way, it is clear that ' because he is blind ' is only 
a sufficient explanation in practice because a pro-attitude 



towards helping the blind is so common that it can be taken as 
understood. Indeed, by omitting to mention it, we show that 
it is so understood; for only so could 'because he is blind' 
serve to explain Jones' action at all. 

It might be thought unnecessary to labour this point; and 
so it would be if some philosophers had not suggested that, 
because statements of fact are commonly given and taken as 
complete explanations, explanations can be given without 
reference to pro-attitudes. Hence they have dismissed as logi- 
cally senseless in certain cases questions in the form ' I know 
that these are the facts, but why did he do it?' and, what is 
more important, 'I know that these are the facts, but why 
should I do it ? '. But these questions are otiose in practice only 
because a pro-attitude of the agent towards doing his duty 
can, in practice, be assumed. They are not logically senseless 
in the way that it is senseless to drive a wedge between having 
a pro-attitude towards doing something and choosing to 
do it. 

3. A-sentences. We often use A-sentences for the purpose 
of explaining someone's choosing to do what he did and also 
for making up our minds what to do. And they are more 
intimately tied to choosing than are ordinary statements of 
fact. Their explanatory function is, as we have seen, partly 
that of showing that the choice was not an unusual one and 
partly that of implying the presence of reasons for choosing 
and indicating of what general type they are without speci- 
fying them. I may choose a book because it is amusing and 
this is not a different reason from choosing it because it has 
a number of good stories or jokes in it. 

In the context of choice the subjective-predictive force is 
usually prominent, because the question whether the book 
would amuse you or anyone else is irrelevant. But it is a mis- 
take to suppose that the subjective-predictive force is essen- 
tial. I may be choosing a book for a Christmas present; and in 
that case the question whether it would amuse me is irrelevant. 

There is an air of tautology about 'I laughed because it 
was amusing', 'I ran away because it was terrifying' and 'I 



remonstrated because it was an objectionable proposal ' which 
is absent from explanations in terms of statements of fact. But 
in spite of this A-sentences do not give complete explanations. 
This is partly because the explanatory force is inversely pro- 
portional to the obviousness of the tautology. ' I objected 
because it was objectionable ' implies that I had some reasons 
for objecting but does not even begin to say what they were. 

And even in the most tautological-seeming cases the A- 
sentence only shows that my choice was not unusual; it does 
not explain why I chose. One often refrains from objecting 
to things that are objectionable. In addition to believing that 
the A-word is applicable, I must have a pro-attitude towards 
the reaction; and, though the reaction may be anybody's, the 
pro-attitude must be mine. My belief that a book will amuse 
or enlighten me (or you) is not a sufficient explanation of my 
choice unless I also have a pro-attitude towards my (or your) 
being amused or enlightened.^ 

If a man were asked why he chose this car rather than that 
he might reply 'because it has more leg-room' or 'because it 
is more comfortable '. The second reply is nearer than the first 
to giving a full reason. 'Why do you choose a comfortable 
car ? ' has an odder ring than ' Why do you choose a car with 
more leg-room?'. But even so, it is not a full reply; and it is 
in some ways less adequate. The descriptive sentence gives 
more information about the car; while the A-sentence con- 
textually implies that it has some of the attributes that make 
cars comfortable without specifying them. On the other hand, 
the A-sentence states what is only implied by the descriptive 
sentence, that the choice turned on the question of superior 
comfort. Neither gives a full explanation since it would not 
be logically odd to ask: 'Why did you want a comfortable, 

I . Failure to distinguish between the reaction concerned (which may 
be anybody's) and the pro-attitude towards it (which must be the 
agent's) is partly responsible for the philosophical muddle over ' ego- 
ism'. Your being hungry may, in one sense, be the reason for my 
sharing the last crust with you; but, in another sense, the reason is 
that I want to relieve your distress. 



This point is obscured by the fact that A-words can acquire 
a pro- or con-force because people normally have only one of 
these attitudes to the reaction concerned. Thus because people, 
on the whole, like being amused and dislike being bored, 
'because it is amusing (boring)' may be given and taken as 
a complete reason for reading (refusing to read) a book. But 
this only applies if and when the contextual background in- 
cludes the relevant attitude. It would be excessively tedious if 
we had to give full explanations on all occasions. 

4. G-sentences, 'ought', an4' right'. The role of G-sentences 
must be examined at length later; here it is only necessary to 
remark that 'because I thought it right (my duty)' is never, 
by itself, a complete explanation of why someone does some- 
thing; nor is it a logically complete reason for choosing. If 
* it was the right thing to do ' and ' it was my duty ' are taken 
to be statements of (non-natural) fact, they suffer from the 
same logical incompleteness as explanations of action as do 
all other statements of fact. Before they can be invoked as 
explanations they must at least be transformed into 'because 
I thought (or knew or believed) it was my duty'; and even 
this will not do. For a gerundive belief no more explains an 
action than any other belief unless ' and I had a pro-attitude 
towards doing my duty ' is added. 

In practice this addition is never made because it is con- 
textually implied. For the phrase ' I thought it was my duty ' 
is a Janus- phrase which we use not just to express a belief but 
to express a special pro-attitude towards doing something, 
namely the 'sense of duty'. It is logically odd to ask a man 
why he does something when he has said that he believes it 
to be his duty, because this phrase contextually implies that 
he has a pro-attitude towards doing his duty. Since there are 
no logical limits to the possible objects of pro-attitudes there 
is nothing absurd about having a pro-attitude towards doing 
one's duty as such. 

The same applies to explanations of conduct in terms of 
fittingness or obedience to a moral rule. We often explain 
both why we did something and why we intend to do some- 



thing by saying that it is fitting or appropriate to a situation 
or that it is in accordance with a moral rule. But these explana- 
tions are only logically complete if they contextually imply 
a pro-attitude to doing what is fitting or to obeying the rule; 
and this, in practice, they always do. But a part of an explana- 
tion that is so obvious that it can in practice be left out must 
not, on that account, be assumed to be unnecessary. 





In the last chapter I mentioned a number of expressions that 
can be used in such a way that they block any further attempt 
to ask the question ' What shall I do ? ' or the question ' Why 
did you do that?'. And I tried to show that a number of 
possible candidates for this role fail to fulfil it and are only 
given and taken as full explanations in ordinary life because 
of their contextual implications. 

In this chapter I shall examine a number of explanations 
that are altogether different in form. These are explanations 
which refer, not to the facts of the case, to the circumstances 
or characteristics of the circumstances, but to the regularity 
of the agent's behaviour, to what may be called broadly his 
'disposition' to act in that way in those circumstances. And 
here again we shall find that some dispositional accounts fill 
the bill while others do not. 

(i) Explanation in terms of habit. As usual the most lowly 
type of dispositional explanation is one that tells us little, if 
anything, more than that the action was only to be expected. 
'It's what he always does in those circumstances'. Such explana- 
tions differ from explanations in terms of physical laws or 
reflexes only in that habits are acquired and can sometimes 
be abandoned. The similarity between habits and behaviour 
as a natural object, behaviour that can be subsumed under 
laws which also apply to things, is marked by our calling 
habits ' second nature ' ; and the fact that such explanations can 
be complete - (so far as they go ; for there is a sense in which 
no mere generalization explains) - without reference to a 
pro-attitude constitutes no exception to the rule. For a man 
whose action is explained in terms of a habit does not choose 



to do what he does. He may have deliberately acquired the 
habit; but that is another matter. I may have a reason or 
motive for putting on my trousers in the morning by reference 
to which I could defend my action as a proper one, for example 
that I want to keep warm or do not wish to outrage the con- 
ventions of society; but these are not explanations of why I 
did it this morning. I did it from force of habit. 

(2) Explanations in terms of character. These resemble 
explanations in terms of habit in that the because-clause 
asserts not a cause (antecedent event) but a uniformity; but 
the uniformity is not in this case of the single-track or rela- 
tively restricted type. In the case of reflex action both stimulus 
and response are relatively specific. In the case of habits they 
are wider; and when an action is explained in terms of char- 
acter they are wider still. If we say that Jones helped the blind 
man across the road because he is a kindly person, we are 
committed to more than the assertion that he always helps 
blind men across the road or even that he always helps those 
in distress. Our hearer is entitled to expect to find Jones in- 
quiring after the health of old ladies and giving lollipops to 
children. Though there is no definite range of things that he 
must do to count as kindly, he must not omit too many of them. 

Character-explanations differ from habit-explanations in 
other ways also ; but they have one important point in common. 
You could not explain a man's behaving as he does by saying 
that he is kindly, honest, or courageous unless he regularly 
does kindly, honest, or courageous things. The most important 
respect in which they differ from habit-explanations is that, 
while habit-explanations conflict with choice - a man does not 
choose to do what he does from force of habit - character- 
explanations do not. To say that a man did something because 
he is a kindly person is compatible with, but does not entail, 
his having chosen to do it. The names of virtues and vices 
are, as we shall see, character-words, and explanations of 
action in terms of these traits of character lie half-way between 
habit-explanations and the motive-explanations to be con- 
sidered in the next section. 



Character-explanations, therefore, do not constitute a dis- 
tinct, class. Their function is to explain an action without our 
committing ourselves either to asserting or denying that the 
action was one that the agent chose to do. A man may act 
from habitual honesty or because he decides to be honest, and 
to say that he told the truth on some occasion because he is an 
honest man is not to commit oneself to saying whether his 
truth-telling was chosen or habitual. The crucial contrast is 
between habit- and motive-explanations; and I shall now try 
to show that, while habit-explanations are logically complete 
since they do not require any reference to a pro-attitude, 
motive-explanations are only complete when such a reference 
is contextually implied. This is an important point since 
attempts have been made to construe motive-explanations in 
terms of a disposition or tendency to act in a certain way and 
thus to make them analogous to habit-explanations. 



In ordinary life to ask what someone's motive in doing some- 
thing was is usually to imply that the motive was a disreput- 
able one; but this must not blind us to the fact that everything 
which a man chooses to do is susceptible of a motive-explana- 
tion. A motiveless action or 'acte gratuit' is logically impos- 
sible; for it is not something that a man could be said to 
'decide' or 'choose' to do and so would not count as an 
'action'. It is, of course, possible to do something apparently 
aimless, for example for the purpose of proving that a motive- 
less action is possible; but this is foolish, since the action in 
fact has a motive, namely to prove the possibility of motiveless 

There is no special class of motive-words. Motives are 
often referred to by means of abstract nouns such as love, 
hatred, fear, greed, ambition, and jealousy; but the correspond- 



ing adjectives are often used to give habit-explanations and 
therefore not to impute a motive. Motive-explanations are 
usually given only in the case of characteristic and therefore 
regular conduct; but this is not necessarily so. A man cannot 
be called 'generous' unless he acts generously fairly often; 
but he can act once and once only on a generous impulse and 
we could then say that his motive was generosity, however 

I shall now try to show that motive-explanations cannot be 
analysed wholly in terms of dispositions or habits or tendencies 
to act in a certain way; and to do this it is necessary to say 
something about the general distinction between explanations 
in terms of occurrences and explanations in terms of 

If we wish to explain why a piece of glass broke we may 
either say that it was struck with a certain force or that it was ' 
brittle. The first explanation refers to an occurrence which 
was the antecedent cause of the breakage ; but the second does 
not. To say that the glass is brittle is to assert what Professor 
Ryle calls a "law-like hypothetical proposition" to the effect 
that "the glass, if sharply struck or twisted etc., would not 
dissolve or stretch or evaporate but fly into fragments". Ryle 
then claims that motive-explanations are of this second, dis- 
positional type. A motive is not an event or force inside you 
which functions as an antecedent cause; but is a disposition 
or tendency to behave in a certain way when certain events 
occur. There are, of course, many diflFerent types of disposition, 
for example habits, capacities, skills, and propensities; so that 
to call a motive a 'disposition' is not to confuse it with dis- 
positions of other types. The cardinal points of the theory 
are that a motive-explanation can always be translated into 
a hypothetical or more or less complicated set of hypotheticals, 
and that the occurrences mentioned in the if-clauses of these 
hypotheticals must all be public, witnessable occurrences. 

Now it is true that many of the words that we use to explain 

I. The contrast is treated at length by Professor G. Ryle: The Con- 
cept of Mind, chapters 4 and 5. 



conduct do operate in just this way, and some of the words 
that we use to give motive-explanations are also used to give 
purely dispositional explanations. But it does not follow that 
the two types of explanation are the same ; for the same words 
can be used to give explanations of different types. 

Consider, for example, the word 'patriotism'. Ryle is mainly 
concerned to refute the theory that a motive-explanation must 
necessarily refer to sensations (whether genuine or of the 
kind that philosophers sometimes postulate). In the case of 
patriotism he admits that there are sensations, sv^^ellings of 
the bosom, and sinkings of the heart, which might be called 
'feelings of patriotism'; but he denies that an explanation in 
terms of patriotism must necessarily refer to these and, what 
is more important, that these feelings can be construed as 
internal prods which cause the patriotic man to behave as he 
does. Both these contentions seem to me to be correct. We 
decide whether a man is patriotic or not by what he does, not 
by what he feels ; and even if he does feel his bosom swell this 
feeling is not an antecedent cause of his behaviour. 

But what follows from this is not that patriotism is a dis- 
position or that to explain someone's action as being due to 
his patriotism is to give a dispositional explanation. It may 
be to do this; but it may also be to do something else. For, 
granted that he acts in these ways, it is still open to us to ask 
what is his motive for so acting. His actions are quite con- 
sistent with his wanting to gain kudos or his having his eye 
on the post-war political scene; and they are also consistent 
with his wanting to help his country. And it is only in this last 
case that we should call him 'truly patriotic', since patriotism 
consists in doing things for the sake of one's country. As 
Aristotle would have said, this 'for the sake of clause is 
part of the essence of every motive, and it is just this clause 
that distinguishes a motive-explanation from a dispositional 

The attempt to construe motives as dispositions fails also 
for another reason. It is an essential point in this theory that 
the protasis clause of the law-like hypothetical should refer, 



not to some internal event, but to something that happens to 
a man. Ryle adduces the example of politeness. "For example, 
a man passes his neighbour the salt from politeness; but his 
politeness is merely his inclination to pass the salt when it is 
wanted, as well as to perform a thousand other courtesies of 
the same general kind. So besides the question ' for what reason 
did he pass the salt?' there is the quite different question 
'what made him pass the salt at that moment to that neigh- 
bour?'. This question is probably answered by 'he heard 
his neighbour ask for it' or 'he noticed his neighbour's eye 
wandering over the table', or something of the sort."^ 

Now 'politeness' can, but need not be, a motive word. A 
man can be polite from force of habit, and in this case Ryle's 
account is correct; the force of 'he acted from politeness' is 
merely that of relating what he did on this occasion to his 
ways of behaving on countless similar occasions. But ' polite- 
ness ' is not a typical motive-word and is, indeed, not normally 
used as a motive-word at all. If we wish to add that he wanted 
to be of service to his neighbour (which would be necessary 
to make our explanation a motive one), we should tend to call 
it 'kindness', 'thoughtfulness', or 'considerateness'.^ 


But it would still be possible to construe motives as disposi- 
tions if a purely dispositional account could be given of enjoy- 
ing, wanting, and the other pro-attitudes to one of which every 

1. Op. cit., p. 113. 

2. The same may be said of another of Ryle's examples, 'vanity' 
(op. cit., p. 86). On page 325 Ryle says: "We know quite well why the 
heroine took one of her morning letters to read in solitude, for the 
novelist gives us the required explanation. The heroine recognized her 
lover's handwriting on the envelope." But, although the novelist does 
not tell us so, it is logically necessary to add "and she wanted to read 
her lover's letter". This is necessary to complete the explanation; but 
novels would be excessively tedious if novelists did not take this point 
for granted. 



motive-explanation must refer. I shall now try to show that 
pro- and con-words cannot be construed in a purely disposi- 
tional way, although they have dispositional uses and in some 
of them the dispositional element is more prominent than in 

Ryle offers a dispositional account of enjoying which runs 
as follows. "To enjoy doing something, to want to do it and 
not to want to do anything else are different ways of phrasing 
the same thing. ... To say that a person has been enjoying 
digging ... is to say that he dug with his whole heart in his 
task, i.e. that he dug, wanting to dig and not wanting to do 
anything else (or nothing) instead. His digging was a propen- 
sity-fulfilment." "Delight, amusement, etc. are moods, and 
moods are not feelings." "In ascribing a specific motive to a 
person we are describing the sorts of things that he tends to 
try to do or bring about. "^ 

Later Ryle distinguishes enjoying from other types of dis- 
position, such as habits and professions, on the grounds that 
a person can only be said to 'enjoy' an activity if he tends 
to be wrapped up in daydreams and memories of it and to 
be absorbed in conversation and books about it. And the 
journalist who is keen on fishing "does not have to try to 
concentrate on fishing as he has to try to concentrate on 
speeches. He concentrates without trying. This is a large part 
of what 'keen on' means. "^ 

This analysis of the differences between what we might call 
'pro-dispositions' and other dispositions requires a reference 
to being 'wrapped up in', being 'absorbed in', and 'trying'. 
And all of these imply pro-attitudes; but it is not clear that they 
are all dispositional. Many pro-words, 'enjoy' for example, 
have dispositional uses. A man can be said to enjoy music even 
if he is not now listening to a concert. But this entails more 
than that he should frequently listen to music when he has no 
duties and be annoyed at interruptions while listening. He 
must enjoy it in the non-dispositional sense in which he can 
be said to be at this moment enjoying listening to Bach. A 
I. Op. cit., pp. 108-12. 2. Op. cit., p. 133. 



man could not be said to enjoy or to be fond of music if he 
only once enjoyed a concert; but he could be said to have 
enjoyed the concert if he never enjoyed any other music before 
or after. Some pro-words are always used in a dispositional 
way, for example being 'fond of, 'keen on', or 'interested in'; 
but even these entail enjoying the activity concerned in the 
non-dispositional sense. 

There seem to be two main reasons for trying to construe 
'enjoying' in a purely dispositional way. (a) In the first place 
enjoyment is not an episode or occurrence. No one would call 
a man's enjoying a hot bath an 'episode in his career'; but 
the argument from the ordinary use of language is double- 
edged, and no one would call it a 'disposition' either. No 
doubt, if a man wallows long and frequently in hot baths when 
he is not compelled or obliged to do so, we are entitled to 
conclude that he enjoys them; but the conclusion goes beyond 
the evidence. And this is shown by the fact that unphiloso- 
phical people are more inclined to say that ' Only I can know 
whether I am really enjoying something' is more of a truism 
than an absurdity, a point to which I shall return later. The 
disposition-occurrence contrast is a piece of logical termino- 
logy tailor-made to elucidate the explanations that we give 
of non-human phenomena such as the brittleness of glass or 
the solubility of sugar. It is most valuable in these cases since 
it saves us from any temptation we might have to think of 
brittleness or solubility as antecedent causes. And it performs 
a similar service in connexion with many explanations of human 
conduct, for example habits, skills, and competences. But 
there is no reason to suppose that it will elucidate all our talk 
about men. 

(b) The most important reason for the attempt is im- 
doubtedly the belief that, if we cannot translate sentences 
about enjoyment into hypothetical in which the if-clauses 
refer to publicly observable events and the then-clauses to 
publicly observable behaviour, we shall be committed to a 
belief in a Private World in which events occur that can only 
be witnessed by the man 'inside whom' they are going on. 

E 129 


It is impossible for me to go into the controversy about 
introspection here and I can only be dogmatic. There is point 
in such assertions as ' Only I can know that I am in pain, or 
listening to music, or enjoying the music'. Why else should 
doctors often ask us to attend to and make repoi^ts about our 
own pains ? And how could we wonder whether a man is really 
interested or politely dissembling his boredom? A keen eye 
can often detect the stifled yawn ; but there may have been no 
yawn to detect. 

This concession to the Private World theory involves no 
dubious faculty of introspection. From Jones's posture and 
demeanour I cannot tell whether he is listening intently to 
Bach or so stricken with grief that he hasn't heard a note. To 
discover which is the case I have to ask him ; and I trust what 
he says, not because I can correlate his statements with states 
of his mind inaccessible to me, but because I have found him 
to be trustworthy on occasions on which I am in a position to 
check what he says. But Jones's position is different; he can 
tell whether he is listening or not; but he does not need a 
special faculty to reveal this to him. 

If a man claims to know that the dog caught the rabbit, it 
makes sense to ask him how he knows and he may reply 
"because I saw it"; and the notion of introspection as an 
internal faculty arises partly from supposing that the question 
"How do you know that you saw it?" requires to be answered 
in the same sort of way. But all that is necessary for him to 
be able to substantiate his claim is that he should have seen 
it and that he should know the English language. In the same 
way it makes sense to ask a man how he knows that the violins 
played sharp, but if he answers "because I heard them", it 
makes no sense to ask how he knows that he heard them. 
Nor does it make sense to ask a man how he knows that he is 
in pain. He feels the pain and knows that *I am in pain' is 
the proper way to describe his feeling; he neither has nor 
requires a special faculty, over and above seeing, hearing, or 
feeling, by means of which he knows that he sees, hears, or 
feels. Perception verbs have a question-blocking role similar 



to that of pro-words. Just as the latter block the question 
'Why did you do that?', so the former block the question 
'How do you know?'. We can, however, change the ques- 
tion to ' Are you sure ? '. 

Nevertheless, although a man cannot see or hear or in any 
way 'witness' his own seeing or hearing, he can observe his 
own listening; for 'observe' here means 'attend to'. There are 
four comments that must be made on this situation, (a) This 
sort of observing is not analogous to seeing and is not in- 
fallible, and it therefore gives no support to the searchlight 
theory of introspection, (b) To attend to one's own listening, 
as opposed to attending to what one is listening to, is a rare 
and sophisticated thing to do. (c) A man who does this is 
almost certain to listen less well than a man who attends to 
what he is listening to, since it is notoriously difficult to do 
two things carefully at once, (d) It is logically impossible that 
all our observation should be self-observation of this kind. 
For a man could not pay attention to his own listening to music 
if he was not listening to music, and this entails that he is 
both hearing and paying some attention to the music. I shall 
return to this point in the next chapter since it has an important 
bearing on enjoyment. 

'Enjoyment' is neither the name of a type of disposition nor 
the name of a type of occurrence. It is primarily a pro-word 
the function of which is to block the question 'Why did you 
do that?'. It is not the only pro-word and it would be a mis- 
take to try to explain all pro-attitudes in terms of enjoyment, 
though not nearly such a grave mistake as to try to interpret 
them in terms of desire or pleasure. A more detailed discussion 
of the different types of pro- and con-attitude belongs to 
psychology; but one final caution is necessary. Just as there 
are really no A-words or G-words, so there are really no 
pro-words or con-words. A word has a pro- or con-force when 
it has the force of making any further explanation of conduct 
unnecessary. But almost any word can be used with this force 
if the context implies it; and any word that is normally 
used with a pro- or a con-force can, on occasion, be used 



with no such force or even with the force opposite to its usual 


The logic of language reflects empirical truths that are so 
general and obvious that we can afford to ignore exceptions; 
and most people agree in their attitudes towards certain bodily- 
sensations. This agreement is reflected in the logic of ordinary 
language in two ways. In the first place, words such as 
wounded, gored, mangled, bruised, out-of-joint, frost-bitten, 
pinched, scalded, and burned can be used to describe bodily 
conditions of which a man can be, and in severe cases often is, 
unconscious. They can also be used to describe one's own 
sensations. But, except in the driest of medical contexts, they 
are almost always used with a con-force. So that if we know 
that a man has frost-bite and is not unconscious it is needless 
to add that he is suffering from frost-bite. And ' because I have 
got cramp' is a complete answer to 'why did you move your 

Secondly words such as ' pain ' and ' hurt ' that are normally 
used with a con-force can also be used as generic expressions 
to refer to those sensations and bodily conditions towards 
which most people normally have a con-attitude. They then 
become descripitve, and as far as I can see there is no reason 
why 'enjoy', 'detest', 'like', and 'dislike' should not go the 
same way. It is sophisticated, but not logically impossible, to 
enjoy a painful sensation and to want to prolong it; there are 
no logical limits to the possibilities of masochistic enjoyment. 
But this fact in no way weakens the logical thesis that, for a 
man to explain fully why he does something, he must at some 
point or other use a word with a pro- or a con-force, even 
though his case may be so complicated that he has to make 
use of a whole hierarchy of expressions. 



Egoism and Hedonism 


Ethical theories are often divided into teleological theories, 
according to which the notions of duty, rightness, and obliga- 
tion are supposed to be defined in terms of or in some other 
way dependent on the notions of goodness and purpose, and 
deontological theories, according to which the notion of 
obligation is incapable of being analysed or made dependent 
in this way. We saw in the first two chapters that, while the 
main ethical tradition from Plato onwards has been teleo- 
logical, intuitionist theories are almost always deontological. 
The difference between the intuitionists and their predecessors 
(except Kant) is therefore more profound than might at first 
appear. It is not only that they make moral judgements 
theoretical or that they gave a special account of how we come 
to know moral truths ; these points would be interesting only 
to the professional philosopher. A point of more general 
interest is that they put an emphasis on the Sense of Duty as 
a unique motive which the older theories would not have 
put. Rightness and Obligation are, they say, the central 
concepts of ethics ; goodness and purpose are of minor impor- 

In the third part of this book I shall argue in favour of a 
modified teleofogical theory. I shall not try to define deonto- 
logical words in teleological terms, but try to show that pro- 
words are logically prior to deontological words in a looser 
sense. They form part of the contextual background in which 
alone deontological words can be understood, while the reverse 
is not the case. In this chapter I shall examine one of the 
most important arguments for according such a pre-eminent 
place to the sense of duty. This argument is nowhere, so far 



as I know, clearly stated; it seems rather to be an assumption 
that certain writers do not question. 

Kant was the first philosopher to make deontological con- 
cepts central in ethics in a clear and uncompromising way. As 
Professor Paton puts it: "Kant knew, of course, that he was 
trying to do something which no one had succeeded in doing 
before - namely, to set forth the first principles of morality 
apart from all considerations of self-interest."'^ 

The deontological argument starts by assuming that every 
action which is not done from the sense of duty is done from 
'inclination' or from 'self-interest', and it further assumes that 
if an action is done 'from inclination' it is done hi order to 
satisfy this inclination. Moreover an inclination is construed 
in the way that Hobbes construes desires and aversions, as a 
sort of itch or craving that I wish to satisfy. Eighteenth- 
century writers make frequent use of the word ' uneasiness ' in 
this connexion, thus assimilating all voluntary action to that 
of a man who moves his limbs to remove some discomfort. 
Now from these premises it is easy to pass to the conclusion 
that all action, except that motivated by the Sense of Duty, 
is really selfish. Suppose, for example, that I do what everyone 
would call an 'altruistic action', such as giving money to a 
charity or helping a friend in need at great cost to myself and 
without hope of reward; and suppose that I do this, not 
because I think it my duty but because it is just what I want 
to do. Although the action is apparently altruistic, it is easy 
to make out that it is covertly egoistic. For, so the story goes, 
I do not do these things in order to help the recipients of the 
charity or in order to help my friend, but in order to satisfy 
my own desire, to allay my itch, to eliminate my uneasiness. 
Fundamentally I am indifferent to the welfare of others and 
concerned only to promote my own pleasure and minimize my 

I. The Categorical Imperative, p. 15 (my italics). The assumption 
that Duty and Self-interest are the only two motives (or types of mo- 
tive) comes out clearly in the same author's The Moral Law, pp. 14, 
15, 19, 25, 31, 35, 46, and 52. 


own pain ; and surely to be concerned only with this is to be 
selfish ? 

Now, if it were true that all actions except those done from 
the sense of duty were really selfish, it would indeed be neces- 
sary to treat the sense of duty as a special, and as the only 
morally good motive. For two things are quite certain : (a) that 
men do in fact often act altruistically, and (b) that moral codes 
universally advocate altruism and condemn selfishness; and 
this they could hardly do if men were necessarily selfish. But 
in fact the premise on which this argument is based is a 
complete muddle. 


Hedonism. I shall examine the accusation of egoism later ; but 
since its plausibility depends largely on the view that all action 
which is not inspired by the sense of duty is aimed at getting 
pleasure for the agent or avoiding pain, it is necessary first 
to examine hedonism. This is the doctrine that all desire is 
for pleasure, and it has not been held nearly so often as its 
opponents suppose.^ 

For the hedonist is not propounding a psychologist's 
generalization about human nature, but a philosopher's para- 
dox. He seems to be donning the mantle of a Rochefoucauld 
or a Freud and telling us (what we all know) that men mistake 
their own motives. They may think that they desire office or 
riches or the public good; but their real aim is something 
quite else. The only diflFerence - and it is a difference that 
shows that the hedonist is a philosopher, not a psychologist - 
is that he seems to be telling us not that we sometimes mistake 
our aims, but that we always do. WItenever, he seems to say, 

I. It was not, for example, held by Hobbes, who is always called a 
'hedonist'. Hobbes takes the trouble to list a large number of desires, 
for office, riches, etc. Can it seriously be maintained that he is saying 
"Ambition is the desire for office; but don't bother to remember this. 
In fact no one desires office; men only desire pleasure."? 


a man thinks he desires something other than pleasure, he is 
always mistaken. 

But the hedonist is not really concerned to deny the obvious 
fact that people enjoy eating and drinking, playing golf and 
doing crossword puzzles and even, on occasion, helping their 
neighbours. Rightly or wrongly, he is inviting us to construe 
'pleasure' as an internal accusative after the verbs 'desire' and 
'enjoy'. He is like a man who says (paradoxically) that we 
only see colours, when he means (truistically) that everything 
we see is coloured. This thesis is a paradox only because it 
seems to entail that we do not see men and trees and football 
matches. Similarly the truism that bomber pilots only aim at 
targets seems to be, though it in fact is not, incompatible with 
the statement that they aim at docks, factories, and marshalling 
yards. Except when 'target' means 'the round thing you shoot 
at on a practice range', it means 'whatever you aim at'. The 
word 'target' would not appear in the list of targets which a 
squadron-leader gives when he briefs his pilots, and it would 
be absurd for a pilot to say " I can't really aim at that factory, 
because, you know, people can only aim at targets". 

The confusion between objective accusatives after the verbs 
'desire' and 'enjoy' (e.g. enjoy food, climbing, a game of 
cricket, listening to Bach) and the internal accusative (desire 
pleasure, enjoy a good time) has led some philosophers to 
dismiss hedonism as a theory so obviously false that it must 
be a puzzle to know how it ever came to be held. Professor 
Vivas, for example, uses the well-known Paradox of Hedonism 
to refute it. The more you pursue pleasure, the less of it you 
get. "The glutton wolfs the food, the drunkard gulps the 
drink, and thus they put up with stuff of dubious quality. The 
result is ironic, even paradoxical; the man intent on pleasure 
is a poor sensualist and is likely to enjoy less of it than the 
man disinterested enough in pleasure to be able to dis- 
criminate the objective qualities of the objects of apore- 
hension"! v 

This criticism altogether misses the mark; for the hedonist 
I. The Moral Life and the Ethical Life, p. 67. 


will reply - as Mill did - that Mr Vivas equates pleasure 
with gluttony and drunkenness and then contrasts these with 
the 'discrimination' that is the aim of the gourmet. But this, 
in the hedonist's termmology, is to dispute the rival claims of 
the pleasure of getting drunk and the pleasure of discrimina- 
tion. Both the drunkard and the gourmet aim at pleasure ; but 
what pleases each of them is a different thing. It may be an 
empirical truth that the gourmet gets more pleasure than the 
drunkard; but, if so, it is not because he is 'disinterested in 
pleasure ' while the drunkard is not. A man may drink Chateau- 
Laffitte because he enjoys it or because he wants to display his 
wealth or his ability to distinguish it from Chateau- Latour or 
because he wants to get drunk. In the last three cases we 
might say that he drinks it as a means to something else or 
with an ulterior motive. But in the first case he does not; what 
pleases him is the drinking. 

Suppose that a man were to say to himself: "The only thing 
in life worth going for is pleasure; let others waste their time 
reading books, drinking wine, doing crossword puzzles, play- 
ing cricket. These are only means to pleasure. I, for my part, 
will have none of them. I shall go direct for the pleasure." 
Such a man would indeed make the mistake that Mr Vivas and 
many other critics attribute to hedonists and he would cer- 
tainly fail to achieve his object; for it is a logical truth that 
you cannot enjoy the pleasure of playing cricket without play- 
ing cricket. But it is not a mistake that hedonists have either 
made themselves or supposed to be at all comm.on. 

What the hedonist really does is to treat pleasure as a 
common ingredient in all the various things that we find 
pleasant and to say that when a man does something he always 
does it either because he expects it to be pleasant or because 
he believes it to be a means to obtaining something pleasant. 
To desire something is to expect it to be pleasant and to enjoy 
something is to find it pleasant. ' Doing something for its own 
sake' means the same as 'doing something for the sake of 
pleasure'. But this is a mistake, not an empirical mistake about 
the sorts of things that people desire and enjoy, but a logical 



mistake about the use of pro-words. And we must now see 

'Pleasure', like most of the abstract nouns in the philoso- 
pher's vocabulary, does all very well for a chapter heading or 
a short, summarizing phrase. In the body of the chapter we 
should be more concerned with 'pleasing', 'pleasant', 'being 
pleased with', and 'pleasures'. And, if he concerned himself 
with these, it would be immediately apparent that the philo- 
sopher was up to the well-known trick of using in a very 
general way a word or set of words that normally have a much 
more restricted use. These words are certainly pro- words; if 
a man says that he finds doing something pleasant, it is logi- 
cally odd to go on asking why he does it. But 'pleasant' and 
its cognates are not the only pro-words and they do not enter 
into the more serious or important cases of choice. 

There are many things that we do for their own sake, 
among them lying in the sun, doing Symbolic Logic, listen- 
ing to music, and giving a helping hand to a neighbour. 
But only the first (and, in some cases, the third) of these 
would normally be said to be done 'for the sake of pleasure'. 
A man who says that he listens to Beethoven's Grosse Fuge 
or reads Paradise Lost because he finds it pleasant either has 
a lot to learn about music or literature, or, more probably, is 
abusing the English language. The hedonist would have done 
better to express his theory, as Spinoza in fact does, in terms 
of 'enjoyment' ; for this is a word that can be used of the most 
absorbing, profound, and noble pursuits as well as of the most 
trivial. Ordinary folk enjoy doing crossword puzzles; but 
saints also enjoy the beatific vision. 

The antipathy which is felt towards hedonism has its source 
partly in the choice of the word 'pleasure' as the key pro- 
attitude word. People have in fact advocated what we should 
ordinarily call 'a life of pleasure' as the best type of life; but 
the more reputable of the philosophers called 'hedonists' (for 
example, Epicurus, Spinoza, and Mill) have not done so. 
Nevertheless their choice of the word 'pleasure' is, as we have 
seen, unfortunate; and so is the choice of the word 'desire'. 



For this word is most prominently used for the carnal desires 
and it is difficult altogether to free one's mind from the usual 
associations of a word. 

Since carnal desires are desires for pleasant sensations, 
hedonists who represent all motives as 'desires' have misled 
their readers into supposing that, according to their theory, 
the goal of all action must be some delicious sensation. And, 
curiously enough, this has been accepted by many philoso- 
phers as an adequate account of all motives except the sense 
of duty. They have represented these motives as 'desires' or 
'inclinations' and even called them all sensuous; and it is then 
but a short step to regarding them as disreputable or beneath 
the dignity of a rational man. But is the desire to help one's 
neighbour, to feed the poor, or to free slaves a sensuous desire ? 
If so, the word 'sensuous' loses its sting. 

And the hedonist's use of 'pleasure' has led to another un- 
fortunate consequence. It has led philosophers to debate 
whether people enjoy an activity or the pleasure of the activity 
and to treat this question as one that might be settled by 
introspection. Now it makes sense to ask whether a mountain 
climber climbs for the sake of climbing or for the sake of 
getting a view from the top; but it makes no sense to ask 
whether he climbs for the sake of climbing or for the sake of 
the pleasure of climbing. 'Jones enjoys climbing' and 'Jones 
enjoys the pleasure of climbing' mean the same thing except 
that, since 'pleasure' is a narrower word than 'enjoy', the 
latter indicates the general type of enjoyment involved. ('Jones 
enjoys the pleasures of climbing' is, of course, quite different; 
for this implies that he enjoys such things as the view from 
the top, the long cool drink in the pub afterwards, or the 
plaudits of the press.) 

A man who enjoys a bottle of wine, a game of cricket, or a 
symphony does not enjoy his own enjoyment any more than 
a man who observes the antics of a clown observes himself 
observing - though he might do so. But again this would be 
a highly sophisticated thing to do, and, since it is difficult to 
do two things well at the same time, he would almost 



certainly not enjoy the wine, the cricket, or the symphony as 
much as he would if he stuck to business. And again it is 
logically impossible that all our enjoyments should be of this 
type. For one cannot just 'enjoy'; one must enjoy something. 
And if a man enjoys his own enjoyment (where 'enjoyment' 
is not an internal accusative but the object that he enjoys), 
then his enjoyment must be enjoyment of something. A mari 
cannot enjoy his enjoyment of a game of cricket without 
enjoying the game. 


Egoism. Hedonism would not be an important philosophical 
theory if it did not seem to entail egoism, which is either the 
theory that men can only choose to do what is in their own 
interest or that it is only rational to do this. Genuinely altruistic 
action is either a logical impossibility or the prerogative of 
dupes and fools. Now, since it is obvious that moral codes 
commend altruism, we must, if the second version of egoism 
be true, either say that the notion of duty is one that sensible 
men will see through and discard or that doing one's duty is 
only long-term selfishness. And it is because each of these 
alternatives is so unplausible that philosophers have tried to 
show that there is a special motive called the 'Sense of Duty', 
which cannot be treated as a 'desire'. All other motives, it is 
assumed, are desires and the man who acts from them is 
selfish. Thus the protagonists of the sense of duty accept as 
an axiom what is really the most absurd point in their op- 
ponents' theory. 

^ If we remember the confusion between an internal accusa- 
tive and an object, it is not difficult to see how this absurdity 
has come to be accepted. Consider the following conversation. 

A. Why are you buying those apples ? 

B. Because I want to give them to my sick aunt. 

A. Why do you want to do that? 

B. Because I do; that's all. 



A's second question has point, since B's reply to the first 
does not necessarily imply a direct pro-attitude. He might 
have been currying favour, and in that case it would be quite 
proper to suggest that he is 'really' being selfish, because he 
was not really concerned for his aunt's welfare at all. But if 
he accepts B's second reply as the truth there is no further 
room for any questioning. But A is a philosopher and insists 
on probing a little deeper. "You want to," he says; "I sup- 
pose you mean that you like doing that sort of thing." B, not 
being a philosopher, falls straight into the first trap. "Why 
yes," he says, "I suppose I do like doing that sort of thing." 
Sure now of his victim, A continues: "You do it because you 
like it. So what you really wanted was not your aunt's happi- 
ness but your own pleasure." And B falls into the second trap; 
his selfishness is now revealed. 

But A's argument not only reaches a conclusion that we can 
all see to be absurd; it is logically untenable. Paradoxically 
it is in cases where a man is really seeking his own pleasure 
that the absurdity of A's argument comes out most clearly. 
For, if he insists on translating 'I wanted to give happiness 
to my aunt' into 'I wanted the pleasure of giving happiness 
to my aunt', he must also translate the latter into 'I wanted 
the pleasure of the pleasure of giving happiness to my aunt', 
and so on for ever. 

Benevolence, sympathy, and compassion are direct pro- 
attitudes towards the welfare and con-attitudes towards the 
suffering of others. And to explain someone's conduct in 
terms of a pro-attitude is not to lapse into covert egoism. For 
benevolence is not a desire to satisfy my own benevolent 
inclinations or desires, impulses, or itches; it is the desire to 
do good to another; and the accusation of covert egoism rests 
solely on the confusion between the object of desire (wha.t the 
man wants), and the internal accusative after the pro-attitude 
verbs, 'want', 'desire', 'like', and 'enjoy'. This confusion is 
itself due to the philosophical habit of representing all pro- 
attitudes as 'desires'; since, of all these verbs, 'to desire' is 
the one that has the most obvious correlative, namely 



'satisfaction'. But a man does not desire satisfaction in the same 
sense that he desires food or drink or the welfare of his 

If I want to relieve the distress of a beggar, I want to 
relieve the distress of a beggar; I do not want my own happi- 
ness, pleasure, or satisfaction. Likewise if I like giving pleasure 
to my sick aunt, what I like is giving pleasure to my sick 
aunt, not my own liking or even the glow of satisfaction that 
I might get from being benevolent. Indeed, if the latter were 
what I really liked, I should be doomed to eternal disappoint- 
ment, because I could never get just that glow unless I had 
acted for the sake of giving pleasure and not for the sake of 
the glow. 

A man may give sixpence to a beggar for any of the fol- 
lowing reasons: to be seen and applauded by others for his 
generosity, to get the thanks of the beggar, in the hope that 
the bread he cast on the waters would one day return to him, 
to quiet his conscience, or to satisfy his generosity. In each of 
these cases except the last the relieving of the beggar's dis- 
tress might plausibly be represented as a means to the agent's 
own happiness and he might well be accused of covert egoism, 
because in each case he was really indifferent to the beggar's 

It is the last case that is important; for if we put it in this 
way, we are apt to suppose that a man who does something 
'to satisfy his generosity' is concerned, not with the beggar's 
welfare, but with his own satisfaction. But this phrase is a 
misleading way of saying that his motive was generosity and 
this entails that he gave the beggar sixpence in order to relieve 
the beggar's distress. If this too is covert egoism, the accusation 
ahogether loses its sting. It is a tautology that all my desires, 
inclinations, wantings, likings, and enjoyments are mine; but 
it is a plain falsehood that what I desire, like, want, or enjoy 
is necessarily my own pleasure or my own anything else; and 
it is also a plain falsehood that a man who does what he wants 
to do or 'acts from inclination' always acts selfishly. 

To be selfish is not to do what one wants to do or enjoys 



doing, but to be hostile or indifferent to the welfare of others. 
It comes out in two ways, (a) A man whose dominant desires 
were for his own pleasures (in the ordinary, not the philoso- 
pher's sense) and who seldom or never wanted to do good 
to others would be a selfish man. (b) A man who does what 
he wants to do or what he likes, when he does it at the expense 
of others, is a selfish man, even if what he does is not in itself 
selfish. To eat when one is hungry is certainly not altruistic; 
but it is not selfish either. What is selfish is to eat the last 
biscuit when others are hungry too. The man who prefers to 
give the last biscuit to his neighbour to eating it himself, 
whether or not he is morally better than the man who does it 
because he thinks it his duty, is certainly not a selfish man. 

It is therefore quite unnecessary to invoke the Sense of 
Duty to account for the occurrence of altruistic action. Indeed 
the argument that is commonly used to show that altruism is 
disguised selfishness when it is not conscientious altruism can 
be turned against conscientiousness itself. If we are made to 
say that the man who purports to help others from a genuine 
desire for their welfare is really a covert egoist who only 
wants to satisfy his own benevolence, why should we not say 
that the man who purports to help others from a sense of duty 
is really a covert egoist who only wants to satisfy his sense 
of duty ? The argument is as absurd in the one case as in the 

Even if it were true that all non-conscientious actions were 
done in order to satisfy our own desires or inclinations, it 
would not follow that non-conscientious actions would all be 
selfish in any usual sense. For the line between 'selfish' and 
'altruistic' could still be drawn in the place that we habitually 
draw it. The diflFerence between the egoist and the altruist 
would now lie in the fact that the former is inclined or desires 
to do whatever benefits him at the cost of his neighbours, 
while the latter is inclined or desires to do what benefits his 
neighbour even when it is to his own disadvantage. And if 
both these types of action are called 'selfish' or 'egoistic' in 
some esoteric, philosophical sense, it is no longer obvious that 



we can only account for self-sacrifice by introducing the sense 
of duty as a special motive which alone is not 'selfish' in this 
strange sense. And, as we shall see, this device does not even 
do the job assigned to it; since the man who helps others 
because he thinks it his duty to do so is not an altruist. He 
may or may not be a morally better man than the altruist, but 
that is another matter. 



Advice and Exhortation 


Practical language is not only used for making up our own 
minds what to do ; it is also used for telling others what to do, 
and there are four main types of situation in which we do 

(a) Cases of giving instructions about the best, simplest, 
most convenient, etc., way of doing what the recipient of the 
instructions has already decided to do or of achieving an end 
that he has already decided to aim at. These cases do not seem 
to give rise to any philosophical difficulties since, although 
such instructions cannot be identified with hypothetical state- 
ments in the form 'You will only succeed if you do . . . ', their 
value as instructions depends mainly on the truth of the hypo- 
theticals with which they are intimately connected; and this 
is an empirical matter. 

(b) Cases in which Jones tries to help Smith to solve a 
problem of choice, which is not just a problem about means 
or methods. I shall call these cases of Advice. 

(c) Cases in which Jones tries to persuade Smith to do 
something by using the language of advice without the proper 
contextual implications. I shall call these cases of Exhortation. 
The choice of the words 'advice' and 'exhortation' to dis- 
tinguish the two types of case is somewhat arbitrary; but the 
distinction between the cases is real and important. 

(d) Cases in which Jones commands or orders Smith to do 

In this chapter I shall consider cases of advice and ex- 

Advice. It is of the essence of 'advice', as here used, that 
if Smith accepts Jones's advice he chooses to accept it. Advice 


is addressed to a free, rational agent who can accept or reject 
it and who must, if he accepts it, have a pro-attitude towards 
carrying it out, since his acceptance is a decision on his part 
to do something. It is not, of course, necessary that he should 
have any pro-attitude prior to the advice being given; Jones 
may advise him to do something that it had never occurred 
to him to do. But it must be something towards which he has 
a pro-attitude now that he considers it. 

Sentences which are not, on the face of them, hortatory but 
are normally used to state facts, can be used to give advice; 
but it is clear that they contextually imply more than they 
state. If Smith asks 'Shall I take an umbrella?', Jones may 
reply 'I don't think it's likely to rain'; bvit this only counts 
as advice if it carries certain contextual implications which I 
shall examine in connexion with A-sentences. 

A-sentences in the context of advice. The situation is radically 
different from that of choosing, since there are now two people 
involved and we must consider each of their points of view. 
Since it must be presumed that men intend (and therefore 
wish for) the natural consequences of their actions, the giving 
of advice contextually implies a desire on the part of the giver 
that the advice should be accepted. It would be logically odd 
to say 'You had better do this, but I hope you won't'. We 
shall see that this does not imply that advice can never be 
given disinterestedly. Jones must hope that his advice will be 
taken, but he need not expect to gain by this. 

Secondly, if Jones uses an A-sentence, part of its force will 
be predictive. The reaction predicted may be anyone's, but 
the pro-attitude towards this reaction must be Smith's, since 
it is Smith's problem that he is trying to solve and to this 
Smith's pro-attitudes are alone relevant. 

Once again, it is clear that when an A-sentence is used to 
give advice it cannot function as an extra premise linking a 
statement of fact to the decision to accept the advice. Smith's 
problem is, for example, whether or not to go to the Super 
cinema this evening. As we have seen, his problem is solved 



once he has been given a good reason for going, provided that 
there are no counter-reasons; and, if Jones's advice takes the 
form of an A-sentence, it must contextually imply that Smith 
has a pro-attitude towards the connected reaction. Thus if 
Jones says that there is an amusing or instructive film at the 
Super this can only count as advice if it is assumed in the 
context that Smith's problem is how to amuse or instruct 
himself. And the fact that other people have been amused by 
the film is not, by itself, a good reason for Smith's going to 
it. The only good reason is that he himself will be amused, 
and this is reached by an inductive step from 'many people 
have been amused'. If we try to construe ' amusingness ' or 
' instructiveness ' as objective properties that do or do not 
reside in the film. Smith might well ask what reason there is 
for supposing that because the film has this property he will 
enjoy it, and that is what he wants to know. 

It may sound absurd to say "I know it is enjoyable, but 
will I enjoy it?" and also absurd to say "I know most people 
have enjoyed it, but is it enjoyable?"; but these cannot be 
simultaneously absurd, for it is not absurd to say "I know 
most people have enjoyed this film, but I don't think I shall". 
The role of A-sentences is not to refer to a property or to 
convert an inductive argument into a deductive one. The 
sentence is used in this context as a piece of advice and there- 
fore constitutes a solution to Smith's problem, not something 
from which a solution could be deduced. But it differs from 
'If I were you, I should go to the Super' and from 'Go to 
the Super' in that it contextually implies that Jones has good 
reasons for giving the advice, and this is but weakly implied 
or not implied at all in the other formulae. 

Let us assume that Smith's problem is whether to go to the 
Super or not and that Jones knows this. Other assumptions 
are possible and they modify the analysis in various ways. 
But it is essential to consider a concrete case and in all cases 
of advice some assumptions must be gatherable from the back- 
ground ; otherwise Jones does not know what Smith's problem 
is and he is not competent to give advice. If Jones says there 



is an entertaining film on he contextually implies the following 

(a) You are trying to decide whether to go to the Super or 
not and your decision turns wholly on whether you will be 
entertained by the film. 

(b) I believe that you will be entertained. 

(c) I believe this because I know you like thrillers and that 
the film is a thriller. 

It is not suggested that Jones goes through any such train 
of thought or that this is what his reply ' really means ' ; still 
less that he states all this. He has obviously not, for example, 
stated the reasons on which he bases his advice. But all these 
points are contextually implied in that, if any of them is 
absent, Jones is not in a position to give definite advice and 
ought to reply more warily. If, for example, he does not know 
Smith's taste in films, he ought to reply, " It's a thriller and 
was very well reviewed", thus giving conditional advice only. 
The implication would then be, not "Go to the film", but 
"Go, if you like thrillers". If he happens to know that Smith 
is bored by thrillers, he certainly ought not to say that the 
film is 'entertaining', even though he would quite properly 
say this to someone else; for it is Smith's problem that he is 
trying to solve and Smith Would be justly indignant if the 
truth came out. The predictive force of 'entertaining' must 
be that Smith is likely to be entertained; for in this context 
it is the only thing worth predicting. Jones's reply is not a 
'statement of fact' which can be extracted from its context 
and pronounced true or false; it is a piece of advice that 
depends for its interpretation on its context. If the right con- 
textual elements are not present, it is deceptive, inappropriate, 
or irrelevant. 

As in the case of choice, the reaction predicted need not be 
that of the recipient of the advice, but the pro-attitude towards 
it, the desire that it should occur, must be ; for it is his problem 
that is being considered. If Smith asks whether or not Charley's 
Aunt is an amusing play it may be clear from the context that 



he is wondering whether it will amuse his children; his own 
reaction is irrelevant. But Jones's affirmative reply is improper 
unless he believes that Smith's children will be amused; and 
it is assumed that Smith has a pro-attitude towards his 
children's amusement. 

This shows the futility of trying to give any general analysis 
of what is meant by ' X is amusing ' divorced from its context. 
Does it mean that all or most people will be amused or that 
all people of a certain type will be amused ? It means that the 
relevant person will be amused and this depends in each case 
on the context. This is obvious in the case of A- words; but 
we shall see that it also appUes to G-words and deontological 
words, which is less obvious and more important. 

It might be argued that the suggestion that an A-sentence 
combines a generalizing element (People like you have been 
amused) with a predictive element (You are likely to be 
amused) simply slurs over the gap which the consequential 
property theory tried to bridge. But this objection neglects 
the important distinction between logical and contextual 
implication. It is not impossible that the generalizing element 
should be true and the prediction false. And since this is so, 
we cannot say that the A-sentence either means or entails both 
of these. For, if it did, it could only be properly used in cases 
where there is a necessary connexion between the two ele- 
ments. And since this never occurs, we should have no use for 
A-sentences at all. Such a demand would set the criteria for 
their use so high that they could (logically) never be fulfilled. 
The function of an A-sentence is not that of combining a 
prediction and a generalization, still less that of supplying 
some logical cement between the two. It is used to give advice ; 
and its role must therefore be that of predicting the only 
thing worth predicting in the context, namely that which 
Smith is concerned to bring about or avoid. Jones could simply 
say " I bet you (your children) will enjoy it". The special role 
of the A-sentence is to indicate to Smith that the advice has 
some inductive backing. It is used to give advice based on 
reasons without saying what the reasons are, and we can only 



understand its role if we distinguish the purpose for which it 
is used from the conditions which limit its use. 

The complexity of elements in an A-sentence used in the 
context of advice may be brought out by considering what 
would falsify it. Since it is not a statement but a piece of 
advice, we should not normally call it ' false ' at all. We should 
say rather that it was bad advice or that Jones was not in a 
position to give advice. The advice might go wrong in any of 
the following ways. 

(i) Jones may be deliberately deceiving Smith, knowing that 
the film is not to his taste. This is a form of lying. 

(ii) He may have been mistaken in thinking that the film 
was a thriller. But since he did not say that it was he has not 
said anything false. He has misled Smith, but not culpably if 
he had good reasons for supposing it to be a thriller. 

(iii) He may have been careless in supposing that Smith 
would enjoy what he enjoys himself. He said " It is enjoyable", 
when he should have said "I enjoyed it". In this case he has 
abused the conventions under which 'enjoyable' is used in the 
context of advice. 

(iv) He may have given excellent advice ; but Smith's enjoy- 
ment was spoilt by the lady in front who wore a large hat. 
The predictive element was falsified; but he did not give bad 
advice or advise improperly. 

(v) He may not have known what Smith's problem was. 
In this case he was not in a position to give advice and gave 
it improperly, however good the advice may turn out. For his 
undertaking to give advice contextually implies that he believes 
himself to be in a position to give it, and this in turn implies 
that he knows what the problem to be solved is. 

It is not suggested that this is an exhaustive list of the ways 
that advice can go wrong; the argument is designed to show 
that the notion of an objective property of ' entertainingness ' 
which is either present in or absent from the film or present 
to a certain degree prevents our understanding the role of 
A-sentences. For, if there were such a property, Jones's reply 
would be true or false or true up to a point. But the concession 



to complexity involved in ' true up to a point ' is not enough. 
Taken in its context the statement that something is enter- 
taining or boring or frightening contextually implies a number 
of different elements the absence of which can nullify the 
remark in quite different ways. Advice can be misleading, 
inappropriate, unwarranted, disingenuous, improperly given, 
or bad. 


G-sentences m the context of Advice. G-words are those that 
imply not merely that the relevant person is likely to have a 
certain reaction, but that he ought to have it. We have already 
seen that they cannot be sharply distinguished from A-words, 
partly because the same word (e.g. 'eligible') can have an 
A-force in some contexts and a G-force in others, and partly 
because there are some reactions that are so universally en- 
couraged and others that are so universally condemned that 
it is impossible to use the A-word concerned without being 
taken to encourage or condemn the reaction. In default of 
an express withdrawal of the G-force, the A-word always 
carries it. 

I have deliberately postponed all discussion of G-words 
until coming to advice, since their role here is more obvious 
than in cases of choice. This is not to say that we do not use 
G-words or 'ought' when making up our minds what to do, 
but that when we so use them we are, as it were, advising, 
exhorting, or commanding ourselves. If a man prefers one 
thing to another there is no temptation to represent the situa- 
tion as one in which two people participate; but we do this 
quite naturally when a man tells himself that he ought to do 
something. It is an important fact that, while the personifica- 
tion of Desires is always strained and artificial, it has always 
seemed quite natural to represent Conscience as a little voice 
inside me that tells me what I ought to do. And there is, as 
we shall see, an excellent reason for this. 



Just as it is useless to represent A-words as standing for 
objective properties, so it is equally useless to treat G-words 
in the same way. If, for example, there were such a property 
as 'Tightness', it would be sense for a man to say "Yes, I 
know it would be the right thing to do ; but shall I do it ? " 
But this sentence is absurd unless 'shall' has a predictive 
rather than a decision-making role.^ A man who says this is 
wondering whether, when the time comes, he will in fact carry 
out his decision (he might suspect himself of weakness of will) ; 
but he is not making up his mind or trying to solve a practical 
problem. He has solved that when he has decided that it would 
be the right thing to do. 

In the example of the film Jones might have said, not that 
it is entertaining, but that it is worth seeing. The difference 
lies in the fact that, while the A- word is more specific - (it is 
the fact that Smith will be entertained that gives the remark 
its point as advice) - the G-word is less specific but indicates 
more clearly that it is advice which is being given. ' Worth 
seeing' bears on its face a hortatory, commending force that 
* entertaining ' does not. The film may be worth seeing either 
because it is entertaining or because it is instructive ; and each 
of these implies that it has certain causal properties. But the 
gerundive phrase neither states what these properties are nor 
indicates whether they are such as to entertain or to instruct. 
Yet even so it is not a mere prediction; its use contextually 
implies (under Rule 2) that Jones believes it to have some 
of these properties and (under Rule 3) that he believes the 
properties to be relevant to Smith's problem of choice. 

In the same way to say, as the guide books do, that a place 
is ' worth a detour ' is not to say that it has a special, objective 
property of detour-worthiness. It is to advise the reader to 
visit it even at the cost of going out of his way. (Giving such 
advice is what guide books are for.) But a good guide book 
does not give advice irresponsibly. The conventions under 
which they are written, published, and read imply that there 
are reasons for making a detour and that there are a sufficient 
I. See p. 268, below. 


number of people for whom the reasons are good reasons. 
The place has a fine cathedral, a good local wine, or is the 
birth-place of some celebrity. A long guide book will give 
the reasons; a short guide book will not, but their presence 
is implied. 

When reasons are given they may themselves have a gerun- 
dive force without being overtly gerundive in form. Chartres 
has 2i fine cathedral; Dijon supplies Poiilet a la Bresse. (And 
this is not a mere statement of fact; its very mention in 
the guide book implies that it is a dish zcorth eating.) Les 
Invalides contains the tomb of Napoleon (and this makes it 
more interesting than a common graveyard). It is clear that 
the simple objective-subjective dichotomy is powerless to deal 
with the logic of such expressions. Is the author stating that 
he admires Chartres or expressing his admiration, as the sub- 
jectivists say? Nothing could be more absurd. He is probably 
by now profoundly bored by cathedrals; and why should 
Mr Baedeker's tastes be relevant to my problems? Nothing 
could be more absurd - unless it be the suggestion that he 
is stating that the cathedral has a property of visit-worthiness 
or admirability that is mysteriously connected both with 
the fact that others have admired it and that I am 
likely to. 

The function of a guide book is to advise, warn, recom- 
mend, encourage, persuade, and dissuade; and the language 
it uses is tailor-made for just these purposes. Generalities 
drawn from the logic of mathematics or natural science or 
ordinary descriptive discourse are worse than useless for 
throwing light on the logic of these linguistic activities. But 
it does not follow that no generalizing is possible; these 
activities could only be carried on in a world in which certain 
facts obtain, and the logic of the language that we use in 
practical contexts naturally reflects our assumptions about 
these facts. 

The first assumption is that there exists a known com- 
munity of interests among a number of people. If you cannot 
tell Poulet a la Bresse from Spam or think that history is bunk, 



this guide book is not for you, because it has no answers to 
your problems. The second assumption is that you have a 
problem, that you are trying to make up your mind what to 
do. Advice must always be for one thing and against others, 
since it is the answer to the question ' What shall I do ? ' when 
this is addressed to someone else. 

It would appear, then, that there is little to distinguish a 
G-word from an A-word and, in the contexts of choosing and 
advising, this is so; for, in these contexts, A-words are being 
used as G-words and differ from them in being for the most 
part more explicit as to the causal properties contextually 
implied. To say that a book is entertaining is not only to 
advise someone to read it but to begin to indicate the reasons 
for reading it. With G-words on the other hand, although 
reasons are always implied, they are in no way specified. A 
second distinction is that, while A-words are often used in 
non-practical contexts, G-words are only used in practical 
contexts. There is nothing odd about discussing the climba- 
bility of a mountain even when no one concerned is thinking 
about climbing and a man who says that it is climbable is not 
yet urging anyone to climb it. On the other hand the prime 
role of G-sentences is to urge, exhort, command, and advise 
and, while there is nothing odd about ' It's climbable, but I 
don't advise you to climb it' (i), there is something odd 
about ' It's worth climbing, but I don't advise you to climb 
it' (2), or 'You ought to climb it, but I don't advise you 
to' (3). It is worth examining the differences between (i), 
(2), and (3). 

(3) is a contradiction-in-use not unlike ' It's a horse, but 
I don't believe it is'. The speaker is advising both for and 
against in the same breath or, if ' advising ' is held to be too 
weak for ' you- ought ', he is commanding for and advising | 
against in the same breath. j 

The logical oddness of (2) is less great ; indeed it is not I 
logically odd at all if the word * you ' is emphasized ; for the 
speaker is, for some reason, expressly excluding you from the 
general injunction to climb, (i) on the other hand is not 



logically odd at all, for there is nothing unusual about telling 
people that something can be done and not advising them to 
do it. 

This brings out one general difference between A- and 
G-sentences. Since the latter belong par excellence to the realm 
of choosing, advising, and exhorting and since the logic of 
these activities is such that sentences used to play a part in 
them must have a pro- or a con-force, G-sentences are always 
explicitly for or against something. A-sentences, on the other 
hand, are neutral unless the context shows which force they 
have. It may not be clear, for example, whether ' It's a dan- 
gerous (or onerous or responsible) post' constitutes advice 
for or against; but 'it's a post worth having' is not so am- 


The redirection of attitudes. An important factor emerges in 
the context of advice that cannot be present in the context of 
choice. If Smith asks Jones what is the best thing for him to 
do the terms of the problem are set for him (Smith) by his 
actual pro- and con-attitudes towards the different situations 
that different choices would bring about. But the terms of the 
problem are not so set for Jones, since Jones may disapprove 
of Smith's pro- and con-attitudes in a way that Smith himself 
cannot, just because they are his attitudes. Jones's task, qua 
adviser, is to solve Smith's problem of choice, to which Smith's 
attitudes are alone relevant. The fact that Jones approves or 
disapproves is not, by itself, a good reason for Smith's doing 
anything any more than ' because he is blind ' is. 

But the terms of the problem are not set for Jones, since 
Jones has his own problem of whether to accept the role of 
adviser or not. If he disapproves of Smith's attitudes he may 
prefer not to solve Smith's problem but to alter it. Thus if 
Smith asks if the film at the Super is a good one in a context 
where it is clear that his problem is 'Shall I go to it this 

i 155 


evening ? ', Jones may reply " You oughtn't to go to the cinema 
at all; you've got some work to do". If Smith asks him for 
the best way of evading income tax, Jones may prefer to 
redirect Smith's attitudes rather than answer the question. If 
Smith asks whether Higginbottom's Introduction to the Calculus 
is a good book (implying ' Shall I read it ? ') Jones may reply : 
"Are you sure it is worth while trying to learn the Calculus 
at your age?". 

This point has certain important consequences ; but it makes 
little difference to the logic of advice. For Jones is still advising 
Smith, only he has chosen to advise him on a different question 
and one that, perhaps. Smith had never faced. Though the 
discussion has shifted on to different ground, it remains true 
that nothing counts as advice unless it constitutes a solution 
to Smith's problem and that this problem is set by Smith's 
actual pro-attitudes. 

If Jones chooses to re-direct Smith's attitudes he may do 
so in various ways, some of which take the form of drawing 
Smith's attention to other pre-existing attitudes of his. 

(a) He may point out certain probable consequences of 
income tax evasion which Smith failed to consider. In this 
case he does not weaken Smith's desire to evade income tax 
but awakens strong counter-attitudes so that Smith no longer 
asks : ' How can I best evade income tax ? '. 

(b) He may point out that the object of Smith's pro-attitude 
is a special case of something towards which Smith has a 
general con-attitude, and thus weaken or destroy Smith's pro 
attitude to this object. Thus he might point out that income 
tax evasion is a form of cheating, and Smith hates and despises 
cheating. When Smith sees the contemplated course in this 
new light he no longer wants to take it. 

(c) He may exhibit his own con-attitude towards Smith's 
considering the original problem at all without, as in the last 
two cases, giving any reason. Now the mere fact that Jones 
has a con-attitude does not by itself constitute any reason for 
Smith's abandoning his original problem; but the very fact 
that Smith asks Jones (and not someone else) for advice shows 



that he has a general pro-attitude towards doing whatever 
Jones suggests, simply because Jones suggests it. Although 
we do not always have to accept advice, the asking of it implies 
some faith in the wisdom and moraUty of the chosen adviser. 
If Jones advises a course that Smith beHeves to be immoral, 
the effect may be that Smith's faith in Jones as an adviser 
is weakened or destroyed; but it may also be that Smith no 
longer regards the course as immoral; he is persuaded of its 
morality by the very fact that the wise and upright Jones 
approves of it. But it still remains true that, if Smith accepts 
the advice, he must have a pro- attitude towards carrying it 
out, even though this is only a pro- attitude towards doing what 
Jones tells him to do, as such. 

It is sometimes said that the role of moral sentences is 
* persuasive ', that they are used to arouse emotions or attitudes 
or to get people to do things. But although, as we shall see, 
moral language can be used in this way, the theory confuses 
the job that moral sentences are used for with the ulterior 
purposes that we may have in using them. Influencing and 
persuading are things that we can do with or without words, 
and their importance for the logic of moral language has been 
seriously overestimated. A man may use advice, as he may 
use bribery, cajolery, or the thumb-screw to persuade some- 
one to do something; but what he actually does with this bit 
of moral language is to advise, not to persuade; just as he 
may use a hammer for making boxes, but what he actually 
does with the hammer is not to make a box but to drive in 

The Persuasive Theory is an incorrect account of the use 
of moral language ; but it enshrines an important truth. Advis- 
ing is something that we choose to do and we must, therefore, 
have some reason for doing it. This reason may be a desire 
to persuade someone to do something and a man who gives 
advice (unless he gives it ironically) must, as we have seen, 
hope that it will be taken. This does not mean, however, that 
advice can never be disinterested. A man who advises another 
on the choice of a career may be concerned solely for the 



welfare of the recipient of his advice, and the father who gives 
death-bed advice to his son can hardly hope to gain by it. The 
Persuasive Theory, by implying that a man who gives advice 
must have an ulterior motive, makes an unfortunate and un- 
necessary concession to the doctrine that all human action is 
necessarily selfish. 


Exhortation. Exhortation is the use of advising language 
without the contextual implication that the recipient has 
some pro-attitude towards adopting the suggested course. 
This might be called ' rhetoric ', ' propaganda ', or ' suggestion '. 
It is a technique much used by evangelical preachers on 
emotional congregations and by advertisers who instil desires 
into their victims by suggesting that they already have them. 
Statements of fact such as ' Spaghetti's, the only Restaurant 
for the real gourmet ' or ' Splosh washes whiter ' are often 
used in this way. 

It is important to notice that this is a secondary use of 
language, parasitic on genuine advice; and this fact is fatal to 
the ' Persuasive ' Theory of moral language. A man can only 
learn to seek, accept, and reject advice if, in the majority of 
cases, accepting the advice does in fact lead to the result he 
himself desires. No one would acquire a pro-attitude towards 
doing what Jones (or everybody) says, as such, if it was not 
on the whole the case that accepting the advice brought about 
what he wished to bring about. But once people have learned 
this, the language customarily used for giving advice can 
operate on them in a more direct way. If the words ' good ', 
' right ', and ' you ought ' had always been used by Jones for the 
purpose of getting Smith to do something which he (Smith) 
did not already want to do, it is inconceivable that Smith 
should ever be moved by Jones's exhortation. It is for this 
reason that the cynical theory that Virtue is the Offspring that 



Flattery begot upon Pride cannot be wholly true.^ Unless 
moral words had first been used in a way which connects them 
with our own interests - whether these be selfish or unselfish - 
we could never have come to be persuaded or dissuaded by 
their use and they could not act, as they sometimes do, as 
levers with which to manipulate the conduct of others. 

I. B. de Mandeville: Enquiry into the Origin of Moral Virtue. Selby- 
Bigge: British Moralists, Vol. II, p. 353. For the criticism see 
Hutcheson: op. cit., Vol. I, p. 81. 





'Good' in the context of choice. We have seen that when 'good' 
is used in the context of choice there can be no logical gap 
between deciding that something is the best or better than its 
rivals and choosing it. This does not imply that there can be 
no discrepancy between the decision which is, on the face of 
it, not a performance of any kind but a judgement, and the 
choice; but it does imply that if there is such a discrepancy 
a special reason must be given for it. And we must now 
consider the role of such expressions as ' because it is a good 
one ' and ' because it is the best ' when they are used to explain 
why a man chose the thing he did. 

The answer to the question ' Why did you choose that car ? ' 
might be a statement of fact (' because it has more leg-room ') 
or an A-sentence ('because it is more comfortable'); and I 
have already discussed the contextual background in which 
such answers can be given and taken as logically complete 
explanations. In each case the car must have some A-property 
and some ordinary, empirical properties on which its A- 
property depends. While the factual answer says what the 
empirical properties are and contextually implies an A- 
property without specifying what it is, the A-sentence does the 
reverse. And each answer implies a pro-attitude towards the 
A-property concerned; otherwise it would not be an answer 
to the question. 

The answer 'because it is the best' functions in a similar 
way, but with certain important differences. In the first place 
it does not just imply a pro-attitude; it expresses it. But it 
does not only do this. If this were all I wanted to do I should 
have to say ' because I happen to like it more than the others '. 

1 60 


It contextual ly implies that I have reasons for my choice; but 
I it does not say what they are and therefore does not explain 
my choice. 

We are tempted to say that it gives the best possible 
reason. After all, what better reason could there be for choos- 
j ing a car than the fact that it was the best available or the best 
I that I could afford? What better reason could there be for 
ji doing anything than the belief that it is the best thing to do ? 
I The trouble is that the reason is too good. It is like saying 
I that I was frightened because it was a terrifying experience; 
I and, as an explanation, it operates in much the same way. 
I Just as ' because it was terrifying ' shows that my fear was not 
j an unusual one and contextually implies that the object had 
j certain unspecified properties by which people are usually 
frightened, so ' because it was the best ' shows that my choice 
was no passing whim, that it was considered more or less care- 
; fully, that the object had certain unspecified 'good-making' 
; properties, and that my choice was not a peculiar one. Any 
of these contextual implications could be expressly with- 
drawn, especially, as we shall see, the last; but in default of 
such withdrawal my audience would be entitled to assume 
them. Just as a G-sentence showed more plainly than an 
' A-sentence that advice was being given but was Jess explicit 
about the reasons, so 'because it was the best' shows more 
plainly that I v/as choosing but says even less about the reasons. 
j In fact it says nothing about them at all; it only implies that 
j I have reasons. The goodness of something is not one of the 
properties for v/hich I choose it. If it were, it would make 
sense to ask why its superior goodness was a reason for 
. choosing it. To ask a man who chose a car because it was faster 
j or more economical or had more leg-room why he chose 
i it is to display ignorance of people's purchasing habits ; to say 
to him " I know you thought it the best car ; but why did 
you choose it?" is logically odd. 

The same logical ties that bind goodness so closely to 
choosing bind it also to activities that are akin to choosing. 
A man who says that he voted for a certain proposal because 

F i6i 


he thought it good has not explained why he voted for it; he 
has merely guarded himself against accusations of flippancy, 
irresponsibility, or indulging in complicated machinations. 
And it is logically odd to say " I think it is an excellent 
proposal, but I shan't vote for it". As we saw, reasons could 
be given for this discrepancy, and the logical nexus between 
thinking good and voting comes out in the fact that we should 
feel entitled to infer that there must be a special reason. To 
call something good is, in a way, already to vote for it, to side 
with it, to let others know where I stand. But it does more 
than this; it implies that I have reasons for casting my vote 
as I do. 

^ Good' in the context of advice. The considerations that 
apply to ' good ' in the context of choice apply equally in the 
context, of advice. And here again the subjectivist is right in 
connecting ' this is good ' with the pro-attitude of the speaker. 
There is the same sort of absurdity in 'This is good, but I 
don't advise you to do it ' as there is in ' This is the best 
course ; but shall I take it ? '. In the latter case the speaker both 
expresses a decision as to how he should act and in the same 
breath asks if he should; and in the former he gives advice 
and in the same breath retracts it. It would be equally odd if 
the hearer were to say "You have told me that it is the best 
course to take; but do you advise me to take it?". 

The differences in the use of ' good ' in advice and choice are 
due to the fact that the problem to be solved is now someone 
else's. The adviser is not making up his mind what to do, but 
helping someone else to make up his mind. And this difference 
brings with it another. The relevant pro-attitude is that of the 
audience. But in other respects the contextual implications 
are the same. To tell someone that something is the best 
thing for him to do is to advise him to do it, but not irre- 
sponsibly. The speaker implies that he has good reasons for 
his advice, that he knows what the problem is and that his 
advice is relevant. The same predictive and causal elements 
are present as in the case of A-sentences ; and advice may, as 
before, be given disingenuously, improperly, mistakenly, or 



unfortunately if one or other of the contextual implications is 


Other uses of 'good'. I shall discuss the other uses of 'good' in 
the order in which they seem to diverge more and more from 
the fundamental use, which is to express or explain a prefer- 

(a) Praising and Applauding. Like choosing, these are per- 
formances, not statements ; and, although in primary uses they 
do express the speaker's pro-attitude, they have other con- 
textual implications which will be examined later. They can 
be done with or without words; but the gestures, hand- 
clapping and the like, which are used for praising have con- 
ventional, symbolic meanings. They mean what they do in 
the way that words mean, not in the way that clouds mean 
rain or cobras in the garden mean trouble. Virtue-words are 
words of praise; and relatively specific words like 'brave', 
'honest', and 'generous' are also descriptive; for they describe 
a person's behaviour and predict the way in which he can be 
relied upon to behave in certain sorts of situation. They both 
praise and give the reason, what the praise is for. But 'good' 
does not do this. In cases where there are recognized stan- 
dards that a man must reach to be worthy of praise they 
contextually imply that he has reached those standards; but 
they do not say what the standards are. ' Because it is a good 
one' does not explain why I praise something; but it does 
imply that the thing has certain unspecified properties for 
which I praise it. My praise was not casual or capricious. 

(b) Commending. The verb 'to commend' is used in two 
ways. It may mean 'entrust to the care of; but this sense is 
irrelevant, since ' good ' is not used to commend in this sense. 
In the sense in which ' good ' is used for commending it is akin 
to praising but has a more hortatory force. To commend 
something to someone is to advise him to choose it. The Oxford 
Dictionary, as we saw, calls ' good ' " the most general adjective 



of commendation " in English; but it goes en to add " implying 
the existence in a high or at least satisfactory degree of 
characteristic qualities which are either admirable in themselves 
or useful for some purpose". 

The form of this definition is interesting, since it brings 
out the difference between the job that the word is used for 
and the conditions limiting its use in a way that philosophers' 
definitions of 'good' never do. The writer of the dictionary 
sees clearly that the word is used to do a job which is not 
'stating' but commending and that the elements of objective 
fact which some philosophers insist on treating as part of its 
meaning are really part of the contextual background of its 
use. In the uses v/hich follow this contextual background 
looms larger and larger, so that in some uses the word ' good ' 
almost comes to be a descriptive word, though, as we shall 
see, it never quite does this and in moral contexts it can 
never wholly lose its gerundive force or its pro-force. 

(c) Verdicts and Appraisals. In chapter i we saw that moral 
language is not only used for choosing and advising, but also 
making moral judgements, which are not decisions to do 
something but verdicts or appraisals of something or some- 
body. Now appraisals are judgements, not just expressions of 
a man's own taste or preference; and it is this point that the 
Consequential Property Theory tries to bring out, but in a 
misleading way. When we judge something to be good we 
always judge it to be good in respect of some property, and 
it is a question of empirical fact whether it has this property 
or not. Thus to judge a wine to be good is not just to express 
a preference for it - and we shall see that it need not be to do 
this at all -; the judgement must be backed by my belief that 
it has a certain bouquet, body, and flavour, and these are 
objective qualities, since a man who found that he disagreed 
markedly from all the experts on these points would admit 
himself to be wrong. It is an essential feature of judgements 
that they are made by reference to standards or criteria; but 
it is necessary to be extremely careful in discussing the way 
in which the criteria are related to the verdict or appraisal. 



Let us assume for the moment that the criteria used by 
experts at wine-tastings, horse-shows, beauty contests, and 
school examinations are agreed to be the proper criteria, 
though this will have to be questioned later. We might be 
tempted to say that if the criteria for being a good X are that 
the X must have properties a, b, and c in some specifiable 
degree, then ' good X ' simply means ' X which has the proper- 
ties a, b, and c in the requisite degree'. But this will not do. 
For it is possible to understand what ' good X ' means without 
knovidng what the criteria are. Thus, if I do hot know the 
criteria used at Crufts I could not tell a good dog (in this 
sense of ' good ') from a bad one or pick out the best dog from a 
group. But this does not mean that I cannot understand what 
' good dog ' means in the way that I could not understand what 
' mangy dog ' meant if I did not know what ' mangy ' meant. 
For I do know that if it is a good dog it must have in a fairly 
high degree those properties which are mentioned in the list 
of criteria for judging dogs, although I do not know what 
these properties are or to what degree a dog must have them 
to rate as ' good '. 

The next two uses are special cases of the appraising use. 

(d) Efficiency. When 'good' is predicated of any object 
(natural or artifixial, animate or inanimate) that is used for 
a purpose it implies the presence in a relatively high degree 
of those properties that the object must have to do its job. 
But again it would be a mistake to say that ' good knife ' just 
means 'knife that is sharp, easily handled, durable, etc.'. The 
connexion between the properties which a knife must have 
to be efficient and its efficiency is an empirical one. We know 
from experience that a knife which has not got these properties 
at all just won't cut and that its relative efficiency at cutting 
depends on the degree to which it has these properties. Nor 
can we even say that 'good knife' means 'knife which cuts 
efficiently', because we could understand what 'good' means 
in the expression 'good knife' without knowing what knives 
were for. But 'good knife' (in this sense of 'good') does 
mean 'Knife which has those properties (whatever they are) 



which a knife must have if it is to do its job efficiently (what- 
ever that is) '. 

(e) Skill. When we call a man a good lawyer, scholar, 
cricketer, or liar, the use is similar to the ' efficiency ' use except 
for the fact that, since these are men, the purpose concerned 
is their purpose, not the purpose they are used for. Just as 
we could not use ' good ' to imply efficiency unless we agreed 
about what the object concerned is for, so we could not use 
it to imply skill unless there was something that was agreed 
to constitute success at the activity concerned. But, just as 
we cannot say that ' good ' means ' efficient ' in the one sense, 
so we cannot say that it means 'successful' in the other. In 
activities involving skill there are rules for achieving suc- 
cess which are stich that we know from experience that unless 
a man applies them he is unlikely to be successful. Thus, if 
we know the rules for success at bridge or cricket we can 
predict, in a very general way, what a good bridge-player or 
cricketer will do; and in calling a man 'good' we imply that 
he applies or follows the rules. This implication can, of course, 
be expressly withdrawn because we know that people some- 
times achieve success in very unorthodox ways. But 'good' 
never quite loses its gerundive force and if we call a man a 
good cricketer without intending to imply that his methods 
ought to be imitated we mislead our audience. 

(f ) The descriptive use. Like most words, ' good ' can be used 
to mean 'what most people would call good'. A man who 
uses it may not be choosing, advising, defending a choice or 
piece of advice, or appraising, but referring to an object which 
he or others would call good if they were doing one of these. 
Thus I may call a wine good even if I am not competent to 
apply the criteria, just because I have heard the experts 
praise it. 

This use belongs to descriptive discourse because it is a 
question of historical fact whether people do or do not call 
the object good, and that is what is being asserted. It is 
necessarily a secondary use, since it would be impossible to 
use 'good' to mean 'what people call good' unless people 

1 66 


called things good in primary ways. And ' good ' is hardly ever 
used with this descriptive force alone. The speaker implies 
that he himself sides with those who call the thing good unless 
this implication is expressly withdrawn or obviously inadmis- 
sible in the context. 


We must now consider the ways in which these uses of ' good ' 
are connected with each other. It is clearly not an accident 
that the same word is used in all these different ways nor 
could this fact be explained in a purely historical or philo- 
logical way. ' Good ' is the Janus-word par excellence ; it is often 
used to do more than one job on one occasion and the logical 
connexions between the various jobs are what they are 
because the facts are what they are. It is also most emphati- 
cally an ordinary, non-technical word and it is a consequence 
of this that the logic of its use reflects empirical truths that 
hold only for the most part and admit of exceptions. For 
ordinary language, unlike mathem.atics, is not deliberately 
constructed by men who have a keen eye for consistency and 
rigour ; it is not deliberately constructed at all but grows and 
changes in an environment in which the exceptional case can 
be and must be ignored. The contextual implications of any 
use of 'good' are many and varied and, on- occasion, any of 
them can be withdrawn, a point which should make us sus- 
picious of counter-examples. It is impossible to understand 
the actual uses of 'good' by considering artificial and excep- 
tional situations because the logic of ordinary language does 
not cater for such situations.^ 

But there is one element v/hich seems to be common to all 
cases. Although a man need have no comparisons in mind 
when he calls something 'good', such comparisons are always 
implied. He must, if challenged, be able to produce examples 
of descriptively similar things that he would call not so good. 
For example, we always praise something with a certain 
I. See pp. 239-44, below. 


degree of warmth which Hes somev/here on a scale between 
mild commendation and hysterical adulation. The word ' good ' 
can be used to express almost any degree of warmth, but it 
must be less than that expressed in the same context by 
'excellent' or 'superb' and greater than that expressed by 
' fair ' or ' tolerable ', 

It is not difficult to understand the connexions between 
the more obviously performatory uses, praising, applauding, 
and commending ; nor is it difficult to appreciate their intimate 
connexion with preference and choice. To praise is not to 
choose; but it is connected with choosing in that it would be 
odd for a man to choose the thing he was prepared to praise 
less highly or not at all. He must have special reasons for this, 
modesty for example, a sense of unworthiness to possess 
the ' better ' thing, or a desire that someone else should have 
it. Again, if a man habitually praises one pianist more highly 
than another we expect to find him attending the recitals of 
the former more regularly and to be more annoyed when he 
is prevented from going. But he might have been told that 
the second is really a better pianist and be trying to cultivate 
a taste for his performance. Explanations can be given of 
discrepancies between praising and choice; but in default of 
an explanation the corinexion is contextually implied. 

If, on a particular occasion, I call a man brave it would be 
logically odd to ask if I was in favour of what he did; for 
'brave' is a praising word and by using it I show that I am 
in favour. Similarly, if I call courage a 'virtue' I show that 
I am, in a general way, in favour of courage, although I might 
not always want to praise a brave deed. It is an empirical fact 
that men are, for the most part, in favour of the modes of 
conduct that they call (descriptively) brave, honest, or 
generous. But this pro-attitude is so widespread that these 
words are not pure descriptive words ; they are terms of praise 
and imply a pro-attitude unless this is expressly withdrawn. 

Now praising and applauding are activities which are often 
performed with the special purpose of encouraging the person 
concerned to continue in the same style, and hissing and 



booing are used with the opposite intention. Although the 
words and gestures employed in praising owe their encourag- 
ing force to convention, they have, granted the convention, 
a natural effect on the people praised.. For it is an empirical 
fact that, except in special circumstances - for example, if the 
praise is considered impertinent - people enjoy being praised 
and are therefore likely to go on doing what they are praised 
for. Praising is logically tied to approval; for if we heard a 
man praise something we could not wonder whether he 
approved of it or not unless we suspected him of being disin- 
genuous or ironical; and it is logically tied in the same way 
to encouraging. But, although it is an empirical fact that men 
tend to encourage and try to promote that of which they 
approve, we must, as always assume that men on the whole 
intend the natural consequences of their actions and therefore 
do not praise that which they would prefer to be otherwise. 
And this assum.ption is reflected in the fact that praising 
implies both approval and encouragement. 

The same logical ties bind praising to advising; it v/ould 
be logically odd to praise one candidate more highly than 
another and to go on to say that one was advising against 
his being given the job or the prize. Odd, but not impossible ; 
for there might, as always, be special reasons for this. 

The "characteristic qualities" which, according to the 
dictionary, are implied by the use of 'good' may be "either 
admirable in themselves or useful for some purpose". In 
contexts involving efficiency or skill it is the latter that we 
have in mind. In such contexts there need be no direct con- 
nexion between the performatory uses, which are all variations 
of ' preferring ' or 'being on the side of, and the usefulness 
implied by 'good'. We may have no pro-attitude whatsoever 
towards the purpose for which something is used or the 
activity at which a man is skilful, as when we speak of a 
'good cosh' or a 'good liar'. But there is still an indirect 
link with the pro-attitudes since 'good' in these contexts 
implies success, and 'success' is a pro-word. A man is not 
a good liar unless he fairly consistently achieves his aim. 



Preference and Appraisal. But it is the connexions between 
the performatory uses and the verdict-giving, judging, or 
appraising use when the qualities on which the verdict is 
based are thought to be " admirable in themselves " that ar€ 
the most important and the most difficult. I shall substitute 
'preferable' for 'admirable', since admiration is itself a per- 
formance akin to praising and 'admirable' is therefore too 
narrow in scope to cover all appraisals other than those of 
efficiency or skill. 

All the performatory uses contextually imply appraisal ; for 
we have seen that it is improper to use 'good', at least in 
an impersonal formula, to express or defend a preference 
unless the preference is a considered one, based on reasons, 
and not unusual. And to say that the preference is ' based on 
reasons ' is to say that the speaker applied criteria or standards. 
It is not necessary that he should have done this deliberately; 
he may have done it automatically; but he must be able to 
defend his choice by an appeal to the standards which justify 

But, although the performatory uses imply appraisal, it is 
not so clear that the converse is true. Indeed it is not true in 
any direct sense; appraisals often imply preference only in a 
roundabout way. For when * good ' is used to give a verdict it 
need neither express nor imply a pro-attitude on the part of 
the speaker. In such cases what a man is primarily doing with 
the word 'good' is applying those standards which are only 
contextually impHed in the more subjective uses. Since ' good ' 
is a Janus-word, he may, of course, be expressing his pre-, 
ferences or advising as well; but he need not be. The embit- 
tered schoolmaster may have no interest in the work of the 
examination candidates at all; he may even prefer stupidity 
to intelligence or have a private belief that the usual criteria 
for intelligence are quite wrong. Nevertheless he may still 
apply the grading words 'good', 'fair', 'poor', and so on in 
accordance with the accepted criteria either from conscientious- 
ness or from habit or from fear of losing his job. 

In the same way a professional taster of wine may dislike 



all wine or prefer the less good to the better; his judgement 
is based solely on the presence of those " characteristic quali- 
ties" which, as an expert, he is able to detect and knows to 
be among the criteria for ' good wine '. But even in these cases 
there is an indirect reference to choosing and advising which 
comes out when we turn from the question "What are the 
criteria in fact used for grading Xs?" to the question "Why 
do we have the criteria that v/e do ? ". Professional wine-tasters 
are, after all, businessmen or the employees of businessmen 
and, though their job may be to taste wine, they only have 
this job because wine is to be bought and sold. It is no acci- 
dent that the criteria for ' good Xs ' are connected with the Xs 
that people prefer or approve of more highly. The profes- 
sional wine-taster may not like Chateau- Laffitte; but he uses 
criteria for judging wine under which it gets high marks 
because people are prepared to pay highly for wine which 
rates highly under these criteria, and they do this because 
they like it. 


Nature and Convention. The dictionary's phrase "admirable 
in themselves" is unfortunately ambiguous. In its context it 
is clear that ' in themselves ' is contrasted with ' for a purpose ', 
and that what the author has in mind is the familiar contrast 
between good-as-means and good-as-an-end. But 'in itself 
is often used in philosophy with at least three other meanings, 
(a) It is sometimes used as a synonym for 'really' or 'objec- 
tively ' to imply independence of human opinion or judgement 
of value. But, in discussing Moore's 'two worlds' argument 
I have already suggested that it is doubtful whether any 
sense can be given to the idea of something being good if there 
was no one to judge it good. 

(b) It is sometimes used with a gerundive force. What is 
admirable or preferable in itself is what people ought to 
admire or prefer. But to use it in this way is not to comment 
on the use of ' good ' but to m.ake a value-judgement, and, if 



the author of the dictionary were thought to be using it in 
this way, he must be thought to subscribe himself to all the 
value-judgements he cites as examples of 'good', (c) But 
'good-in-itself could also be used to mean 'naturally good', 
to inaply that the criteria or standards used for judging the 
goodness of something are not, like the criteria for a good 
postage stamp, dependent on human convention. It is this 
contrast that I propose to discuss. 

We call a taste (or any other pro-attitude) a 'natural' one 
if (a) it is pretty general even among people of very different 
societies and if (b) most people do not have to learn to 
acquire it. It is important to notice that both these criteria 
for what is 'natural' are extremely vague and that they both 
admit of exceptions. A taste for strawberries does not cease 
to be natural because Jones happens not to like them or 
because Smith did not like them at first. Benevolence and love 
of life are natural pro-attitudes, even though there are 
misanthropes and suicides. 

The criteria used for appraising are partly natural and 
^partly conventional. In music, for example, the criteria which 
critics apply to a composition or performance are conven- 
tional in that they vary in different cultures and it is necessary 
to learn what they are; and musical taste is also partly con- 
ventional in that it is not natural to like or admire a Bach 
fugue in the way that it is natural to like sweets or to love one's 
children. It may well be that no criteria or tastes are wholly 
conventional. Correlations can be found between the criteria 
employed and the physiological facts of hearing; for example 
we know that the musical intervals and key-relationships 
on which all western music is based and which enter into 
the criteria used for judging a musical composition are of 
a mathematically simple kind. And even in the case of the 
criteria used for judging dogs at Crufts, which are highly 
artificial, it is possible to trace historical connexions between 
the criteria now used and the criteria that were used when 
dogs were used for practical purposes; and these last were 
natural criteria in that the purposes, such as hunting and 



protection from wild animals, were based on natural pro- 

But in many cases the criteria now used are connected to 
natural criteria only through a long process of change and 
have become modified to such an extent that their original 
connexion with natural pro-attitudes has been entirely lost. 
And in such cases it often happens that we do not use the 
criteria we do because people have the pro-attitudes they 
have, but we have the pro-attitudes we have because the 
criteria are what they are. It may be that no one can now 
remember exactly why certain criteria were originally chosen 
to be the standards of judging something to be good or bad 
of its kind and that people are now prepared to admire, praise, 
and "pay highly for objects- because they conform to the 
accepted criteria, rather than accepting the criteria as ' proper ' 
ones because, under them, the things that they admire rate 
highly. Taste is dictated by fashion, not fashion by taste. 

But such cases must (logically) be secondary cases and it 
would therefore be a mistake to cite them in proof of the 
contention that criteria are logically prior to pro-attitudes. 
For unless there were primary cases in which we adopted 
criteria because we already had a pro-attitude towards the 
objects that in fact rate highly under them, it would be impos- 
sible to understand how the same set of words could be used 
both in applying criteria and for choosing, praising, and advis- 
ing. It is only because ' good ' is used in applying criteria in 
cases where we use the criteria we do because our desires, 
interests, and tastes are what they are that men can come to 
acquire a taste for what counts as 'good' under the accepted 
criteria even in cases where the original connexion between 
the criteria and the taste has been lost. Advertisers and pro- 
pagandists, arbiters of taste and leaders of fashion could not 
(logically) stimulate new tastes and attitudes by the reiterated 
use of criterion-applying language unless this language was 
also used for applying criteria in cases where there are pre- 
existing tastes and attitudes. Without genuine enthusiasts 
there could be no snobs. 



In many cases, therefore, the answer to the question ' Why 
do we, use the criteria for judging so-and-sos that we do?' 
may be of a purely historical kind ; the criteria are traditional ; 
they have been concocted and moulded by interested parties, 
and so on. But this sort of ansv/er cannot be given in all 
cases; there must be some cases in which we use a set of 
criteria because, as an empirical fact, they give higher ratings 
to those objects which we prefer. 

In discussing appraisals I assumed that there was no diffi- 
culty about saying what the proper criteria for judging Xs 
are or about selecting the experts, leaders of fashion, or arbi- 
ters of taste ; and it might seem that these assumptions involve \ 
a vicious circularity in the attempt to construe the grading- 
scale of good, fair, poor, and bad in terms of the standards 
used by experts. But this is not so for two reasons, (a) In 
some cases there are tests of competence which are purely 
objective and empirical. Some men, for example, have perfect 
pitch, can detect minute musical intervals, can recognize and 
accurately reproduce long and complicated tunes, and so on, 
while others cannot ; and these are matters of fact. In judging 
their expertise we must, of course, rely on the ability of other 
experts to assess their competence; but the judgements of 
these experts is 'objective' because they fulfil the require- 
ments for objective language discussed in chapter 4. It is 
possible that one man might have a finer ear than all other 
men, so that in a case in which he said that two notes were 
slightly different when everyone else said they were the same j 
he would be right and they wrong. But if there were no 
indirect tests, such as the appeal to readings of scales and 
meters, for deciding whether he can really detect these dif- \ 
ferences or is only bluffing, and if those who honestly claimed to i 
be able to make fine discriminations did not on the whole agree J 
with each other, we could not call their judgements 'objective', j 

Now from the fact that a man is able to make these fine | 
discriminations or to perform better than others in these I 
objectively testable ways, it does not, of course, follow that 
he is a good judge. For to say that he is a good judge is either 



to State that he is good at applying tlie accepted criteria for 
what is good (which is different from being good at passing 
the objective tests) or to express approval of his judgements, 
to praise him, to encourage others to accept his judgements, 
and so on; and in most cases it is to do all these things at 
once. But, once again, the reason why we allow that, in a 
general wa)^, the most technically competent people are the 
best judges lies in the facts. A man who is tone-deaf is unlikely 
to be able even to distinguish one piece of music from another 
and his value-judgements (if he makes them), are not likely 
to be consistent with each other; so that his value-judgements 
would be useless as a guide to others. A man who knows little 
Greek could not be a good judge of a piece of Greek prose. 
Consistency and fine discrimination are not sufficient con- 
ditions of good taste or moral insight, but they are necessary 
conditions if criteria are to be used for the purposes for which 
they are used. 

(b) Secondly, the person who rejects the criteria usually 
employed or the verdict of the acknov/ledged experts may do 
so in two ways. He may simply refuse to be guided by them 
on the grounds that he happens not to like what is usually 
called good. But, if he goes further and says that the usual 
criteria are not good criteria, he is not just rejecting them; he 
is himself using criterion-applying language and he implies 
that he has second-order criteria forjudging (and condemning) 
the usual first-order criteria. 

To the questioning of criteria there is no end; but if we 
ask whether the criteria for judging Xs are good criteria we 
must, at whatever level we have reached, use criteria for 
deciding whether they are good or not. It is logically absurd 
to ask a question without knowing how the answers to it are 
to be judged to be good or bad answers. The appeal to 
criteria accepted by experts is not circular, but regressive; 
i and the regress is not a vicious one since, although we can 
always question the criteria, there is no practical or logical 
necessity to do so. The self-guaranteeing criteria so vainly 
sought by some moralists are neither possible nor necessary. 



Non-practical appraisals. We often make appraisals in contexts 
where there is clearly no question of choosing or advising, 
for example moral judgements about historical or fictional 
characters. And this seems to involve a difficulty for theories 
which make appraisals logically dependent on pro-attitudes. 
Hutcheson and Hume, for example, tried to reduce moral 
judgements to expressions of feeling. They were not guilty 
of the Naturalistic Fallacy, since they were prepared to 
allow that m.oral approval and sympathy are special, moral 
feelings distinct from other types of feeling. But even this 
concession to the peculiarity of the moral use of language 
does not save them from an important objection that seems 
at first sight fatal to their case. Sentiments, as Hume noticed, 
seem to vary in rough proportion to the propinquity of their 
objects. We are not moved by the iniquity of remote historical 
characters as we are by those closer to us; and we feel more 
approval for and sym.pathy with those near to us than with 
those who are more remote. Yet our moral judgements do 
not vary in the same way. "We read Cicero now without 
emotion, yet we can still judge Verres to be a \dllain. According 
to Hume's theory our judgement must change as do our feel- 
ings. I do not feel indignation as strongly now about the 
German invasion of Czechoslovakia as I did at the time it 
happened; yet I do not judge the action to be less wrong 
than I did then, or the agents less criminal .... It is but 
a v/eak subterfuge to say we transport ourselves by the force 
of imagination into distant ages and countries, and con- 
sider the passions v/hich we should have felt on contemplating 
these characters had we been contemporaries and had com- 
merce with the person. ... I now feel completely indifferent 
to Verres, and know it. Yet, Hume tells me, when I judge 
Verres to have been a villain, I am so deceived by my imagina- 
tion that I talk as if I felt a strong feeling of anger. "^ 

Dr Raphael's criticism is fatal to the theory that a man 
I. D. D. Raphael: The Moral Sejjse, pp. 88 and 91. 


who makes a moral appraisal is always expressing a feeling; 
and a similar criticism could also be made of any theory 
which says that to appraise is alv/ays to praise, advise, com- 
mend, etc. On some occasions a man may be simply applying 
the criteria that he and others customarily use for these 
purposes. To call Verres a villain is to pass a verdict on him, 
to condemn him. Now the Moral Sense School were, I think, 
mistaken in construing moral approval and disapproval as 
feelings, since this suggests too strongly the analogy with 
itches, aches, and tickles. But they were right to connect 
moral appraisals and verdicts with approval and disapproval. 
For although a man who passes a verdict need not be expres- 
sing a pro- or con-attitude, we have seen that the criteria he 
uses are directly or indirectly linked with these attitudes; 
and in the case of moral judgements they must be linked in a 
special way that may be absent in other cases. 

I said earlier that, although in other cases ' good ' might lose 
its gerundive force, it cannot wholly do so when used to make 
moral appraisals. The reason is that, whatever may be the case 
with other types of appraisal, moral appraisals must be univer- 
sal. Anyone who makes a moral appraisal even of a remote 
character must be willing to apply the same criteria univer- 
sally. And it follows from this that he must be willing to apply 
them in practical contexts. If I am not prepared to condemn 
■anyone whose behaviour is like that of Verres in all relevant 
respects, then, in calling Verres a villain, I am not making a 
genuine moral judgement; and the relevant respects are all 
of an empirical, objective kind. It would, of course, be trivial 
to include among them an objective property of villainy or 
moral turpitude; all that is necessary is that I should be pre- 
pared to condemn anyone who did the sort of thing that 
Verres is called a villain for having done, anyone who 
oppressed the poor, robbed the rich, took bribes, and cheated 
the treasury, and all for his own personal profit. 

Moral appraisals are therefore connected with choosing 
and advising in a way that non-moral appraisals need not be. 
It is not logically odd to say " This is the better wine, but I 

G 177 


prefer that"; but it is logically odd to say "This is the 
(morally) better course; but I shall do that''.^ And a man 
cannot be making a genuine moral judgement about Verres 
if he would himself be prepared to act on the same principles 
on which Verres acted and prepared to exhort others to do 
so. In condemning Verres he is not expressing any emotion; 
but he is affirming his own moral principles. 


Objective- Subjective. In chapter 6 I said that the distinction 
between "For what job is the word '. . .' used.?" and "Under 
what conditions is it proper to use that word for that job ? " 
throws light on the objective-subjective dispute. 

As we should expect, both parties are right. Just as the 
subjectivists are right in denying that A-words stand for 
special properties and explaining them in terms of people's 
reactions, so they are also right in connecting 'good' and 
'bad' with people's desires, tastes, interests, approvals, and 
disapprovals. There is a logical absurdity about calling a play 
'amusing' if the speaker believes that it never has amused 
anyone and never will ; and there is the same logical absurdity 
in calling something 'good' without any direct or indirect 
reference to a pro-attitude. If the connexion between ' good ' 
and the pro-attitude that is contextually relevant were not a 
logical one, a gap would emerge between calling something 
good on the one hand and deciding to choose it, choosing it 
or advising others to choose it on the other which v/ould 
make these activities unintelligible. Moreover, the subjecti- 
vists are also right in connecting ' good ' with the pro-attitudes 
of the speaker, at least in moral cases. 

But the objectivists are also right. They are mistaken in 
denying the points made by the subjectivists above and in 
thinking that goodness must be a unique, non-natural 

I. This may sound surprising. We all know what it is to take what 
we know to be the morally worse course. I shall try to remove the air 
of paradox in chapter i8. 



property. It is sometimes argued that if there were no such 
property we could not account for the fact that we use the 
impersonal form 'this is good' rather than the personal form 
'I approve of this', and those who use this argument are 
inclined to forget that we have an impersonal form 'this is 
nice ' as well as the personal form ' I Hke it ', so that niceness 
would have to be an objective property too. 

It would indeed be puzzling to understand why we use these 
impersonal forms if we were just talking about or expressing 
our own approvals ; but this argument does not show that we 
are talking about something else, still less that this must 
be a unique property. We can account for the objective 
formula, as we did in the case of 'nice', by saying (a) that 
'X is good' is not only used in the context of choice and 
(b) that, when it is so used, it implies a great deal that is not 
implied by ' I approve of X ' and is expressly denied by ' I 
happen to approve of X'. It implies that my approval is not 
an unusual one and that I could give reasons for it. It implies 
also - what is a matter of objective fact - that the object con- 
forms to certain standards which are generally accepted. 

It is sometimes argued that * this is good ' cannot just mean 
' I approve of this' on the ground that we can say " I approve 
of this because it is good". Approval must therefore be an 
intellectual emotion which arises in us only when we recog- 
nize something to have the objective property 'goodness'. 
But it has never been clear what the connexion between the 
approval and the recognition of the property is supposed to 
be. Is it logically necessary that anyone who recognizes the 
property should feel approval, or is it just an empirical fact 
that people who notice the property, and only they, have the 
feeling ? Each of these answers involves insuperable difficulties ; 
but if neither is correct we must find some other way of 
explaining the 'because' in 'I approve of X because it is 

The need for such an explanation vanishes when we see 
that this is not a reason-giving 'because' like that in 'I 
approve of Jones because he is kind to children ' but more like 



'I like Jones because he is likeable'. It rebuts the suggestion 
that I just ' happen to ' approve of X and it implies that X has 
certain properties which make it worthy of my approval and 
that it conforms to the known standards for Xs. 

The objectivist is right in drawing attention to the factual 
background which makes impersonal appraisals possible; 
but the facts which it contains are ordinary, empirical facts, 
not special, non-natural facts. Unlike the subjectivist (who 
tends to ignore the background altogether), he tries to 
include the background in the meaning of the word ; and this, 
combined with the mistake of confusing practical and descrip- 
tive discourse, leads him into the vain pursuit of a single 
ingredient to which we always refer when we call something 


The Naturalistic Fallacy. We are now in a position to see why 
the moral philosophers of the past subordinated the critical 
or appraising uses of moral language to the practical uses. 
Each presupposes the other, but in a different way. The 
practical uses presuppose the appraising use in that we could 
not use 'good' as we do for choosing, advising, and praising 
if we did not employ criteria or standards ; since we only use 
'good' for these purposes when we are emplopng standards. 
Nevertheless people who did not know what standards were 
could do things recognizably like what we, who have stan- 
dards, call choosing, advising, and praising. They would be 
very rudimentary performances, hardly deserving the names 
of choice, advice, and praise ; but they could occur. We draw 
a distinction between ' good ' and ' happen to like ' which people 
without standards could not draw; and we, who have the 
distinction, would describe their activities in terms of what 
they ' happen to like ', because they could not do anything that 
we would call 'choosing the best'. In this way the practical 
uses of ' good ' imply the appraising use. 

But the practical uses are logically prior to the appraising 



use in a much more fundamental way. Unless men had pro- 
attitudes, there could not be even rudimentary analogues of 
what we know as appraising, judging, or passing a ver- 
dict. For these involve the use of standards; and without 
pro-attitudes we should neither have any use for standards 
nor even be able to understand what a 'standard' was. We 
can imagine a world in which there was choosing, but no 
appraising and also a world in which there was classifying, 
sorting, and ordering (for example by size) but no choosing; 
but, in a world in which there was no choosing, there could be 
no such thing as appraising or grading. 

Ethical Naturalism is the attempt to trace logical connexions 
between moral appraisals and the actual pro- and con-attitudes 
of men, their desires and aversions, hopes and fears, joys 
and sorrows. One-track naturalistic theories always fail to 
do justice to the complexity both of the facts and of the logical 
connexions, since they suggest that there is only one thing 
towards which men have a pro-attitude, pleasure, or that all 
pro-attitudes are desires. And these theories are both psycho- 
logically and logically misleading. 

Opponents of the Naturalistic Fallacy have pointed out the 
logical errors. It is true that gerundive and deontological 
words cannot be defined in terms of pleasure, desire, or even 
purpose ; and I shall try to show how they are connected with 
these teleological concepts later. It is also true that gerundive 
judgements and value judgements do not follow logically from 
descriptive statements about what men like, enjoy, and approve 
of. But the reason for this is not that gerundive words and 
value words refer to special entities or qualities, but that a 
person who uses them is not, except in certain secondary cases, 
describing anything at all. He is not doing what psychologists 
do, which is to describe, explain, and comment on what people 
like, enjoy, and approve of; and he is not doing what moral 
philosophers do, which is to describe, explain, and comment 
on the way in which people use moral words; he is himself 
using moral language, expressing approval, praising, advising, 
exhorting, commending, or appraising. 



The attack on the Naturalistic Fallacy is thus far justified. 
But the conclusion which is commonly drawn, that moral 
concepts are a special sort of concept which must be purged 
of all association with the ' merely empirical or phenomenal ' 
concepts of enjoying, wanting, and approval is not justified. 
Psychology is not as irrelevant to ethics as some modern 
philosophers insist; for, although moral judgements do not 
follow from psychological statements, we cannot understand 
what the terms used in moral judgements mean unless we 
examine them in the context of their use; and they are used 
either directly to express a pro- or con-attitude or to perform 
some other task which beings who had no pro- or con-attitudes 
could not perform or even understand. The various ways in 
which ' good ' is used are unintelligible unless they are directly 
or indirectly connected with choice; and I shall try to show 
later that the same applies to ' ought '. 

Moral philosophy does not, therefore, "rest on a mistake". 
For the great philosophers were not primarily interested in 
the question whether deontological words could be analysed 
in terms of 'merely empirical' or 'natural' concepts. They 
believed that, human beings being what they are, there are 
certain types of activity that are in fact satisfactory to them 
and that it is possible empirically to discover what these are. 
No doubt they often made mistakes of fact, for example that 
of supposing that what is satisfactory to one man would be 
satisfactory to another; and they made mistakes of logic, for 
example that of supposing that 'good' could be extracted 
from its context and be said to mean the same as ' satisfactory'. 
But they do not seem to have been mistaken in their basic 
assumptions that the language of obligation is intelligible 
only in connexion with the language of purpose and choice, 
that men choose to do what they do because they are what 
they are, and that moral theories which attempt to exclude all 
consideration of human nature as it is do not even begin to 
be moral theories. 


Duty and Purpose 


Right' and 'Ought' 

'Duty' and 'Obligation' are chapter-heading words; I shall 
have more to say about them later; but it will be better to 
begin our discussion of deontological words with the words 
that we actually use when discussing duties in practice, of 
which the most important are 'right' and 'ought'. 

The main purpose for which these words are used is to 
tell someone to do something, but they have subsidiary uses 
not unlike the subsidiary uses of 'good'. They can be used to 
express and to give reasons for verdicts in cases where there 
is no question of anyone's doing anything and even to imply 
no more than that something conforms to an accepted stan- 
dard. But they only come to have these uses because verdicts 
are also made in practical contexts. The objective, fact-stating 
background of their use in practical contexts survives even 
when the context is not a practical one. Thus we can use them 
for passing judgement on a remote historical character or even 
a character in a novel. Henry Crawford ought not to have 
seduced Maria Rushworth. The Revocation of the Edict of 
Nantes was (morally) wrong. Novels and histories are not 
always written with a practical purpose and no one supposes 
that anyone who makes these judgements is advising, warning, 
or commanding. But it would still be wrong to say that he is 
just ' stating a fact ' ; for these are still moral judgements about 
which disputes could arise that could not be settled in the 
same sort of way that a dispute about a date could be settled. 
What he is doing is to condemn; and, although condemnation 
has in this case no practical point, it is difficult to imagine 
how we should ever have come to do such a thing unless we 
did it in practical contexts. 

The difference between 'right' and 'ought' is, roughly, 
that while 'I ought' and 'You ought' are used to express 



decisions and injunctions, 'right' is mainly used to support 
these decisions and injunctions in a special way that must now 
be analysed. 


Right. If someone is asked to explain why he did or intends 
to do something, he may say "because it is the right thing to 
do "; but this explanation has the same near-tautological ring 
as "I was frightened because it was terrifying" or "I decided 
to do it because it seemed to me the best thing to do". It 
would be logically odd for a man to say "I know you thought 
it right, but why did you do it ? " ; but ' I thought it right ' does 
not just express his decision. It indicates the type of reason 
which he had, though it does not give the reason any more 
than 'because I thought it best' does. 

' I thought it right ' is so near to ' I thought it best ' that in 
many contexts there is little difference between them; but 
there often is a difference and sometimes a very great one. 
Since *I thought it right' is tied so closely to the decision to 
act that it is absurd to insert a logical wedge between them, 
its use, as an explanation of choice, contextually implies a 
pro-attitude, which is overtly expressed and not just implied 
by the use of 'good' and 'best'. We have seen that 'good' is 
only used to express approval under certain conditions one 
of which must be that the action or object conforms to certain 
standards. There is a similar limitation in the case of 'right', 
only this time the implication is that the action is fitting or 
appropriate or that it conforms to certain rules. It thus indi- 
cates that the decision is of a different type from that which 
would be explained in terms of something's being good and 
one that is less directly linked to the agent's pro-attitude. 

But, in practical contexts, it must be so linked. Philosophers 
have often attempted to define 'right' in terms of fittingness 
or obedience to a rule; but these attempts fail to do justice 
to the practical role of 'right'. For fittingness is not a two- 


'right' and 'ought' 

term relation that we must 'see' to obtain between an action 
and a set of circumstances. At least one more term is required 
for a full analysis of 'fittingness', if the concept is to be used 
to elucidate 'right', namely the purpose of the person con- 
cerned. This reference to purpose is often so obvious that we 
can afford to suppress it in practice. No doubt the right 
key is simply the one that fits the lock in a straightforward 
spatial sense and the right peg to put in the square hole is 
the square peg. In the same way the right degree of deference 
to pay to a person is the degree which is fitting to his social 

But this two-term fittingness is not enough. In the last 
example there is already a covert reference to a rule of pro- 
priety, a rule which determines what degree of deference befits 
each person ; and in all the examples there is a covert reference 
to the purpose of the person for whom it would be right to 
do something. It is assumed that he wants to open the door 
or to fill the hole exactly. For a man who wants to fool the 
police into thinking that he cannot open the door, the 'fitting' 
key to use is the wrong key. 

The attempts of deontologists to explain 'right' without 
reference to purpose appear satisfactory only because, in some 
cases, a particular purpose is so much more common than any 
other that ' right ' has acquired a purely descriptive force. Thus 
we could either say that the criminal ought to use the wrong 
key (thus tying 'right' analytically to 'fitting' and at the same 
time loosening its ties to 'ought'); or we could say that the 
right key for him to use is the one that does not fit (thus 
tying ' right ' analytically to ' ought ' and loosening its ties with 
'fitting'). And we might even draw attention to the pecu- 
liarity of his purpose by saying paradoxically that the right 
key for him to use is the wrong key. But, however we choose 
to make the point, the fact remains that the key he ought to 
use is not the key that fits. We cannot at the same time tie 
'right' analytically both to 'fitting' and to 'ought' and also 
construe the first of these as descriptive and the second as 
practical. If 'right' is used in such a way that it would be 



logically odd to question its connexion with 'ought' it cannot 
be synonymous with a word of which the function is to refer 
to an objective relation, even a non-natural one. Once more 
the objectivist confuses the job which a word is used to do, 
in this case to explain a decision, and the conditions that must 
obtain for this word to be used, in this case the existence of 
a relation of fittingness. 

The analysis of 'right' in terms of 'in accordance with a 
rule' suffers from a similar defect. The word 'right' is often 
used in a context involving rules or laws and part of its mean- 
ing in such contexts is simply 'in accordance with the rule or 
law'. The rule may either be something that we could point 
to in a book of rules, a volume of Hansard, or a manual of 
etiquette, or it may be, more vaguely, one of the code of rules 
that ordinary, decent people observe. A similar element is 
often present in the meaning of such words as 'fair', 'honour- 
able', 'equitable', 'merit', and 'deserve', and we shall see that 
in attempting to define deontological words in teleological 
terms, teleological writers have not always done justice fo this 
element. But the mere existence of a rule is not a (logically) 
good reason for anyone's doing anything; and if the function 
of 'right' was simply to promulgate or draw attention to a 
rule without implying a pro-attitude towards obeying it, it 
could not be used to give logically impeccable reasons for 
choosing and advising. 

The point is often obscvired by the fact that, since there are 
no logical limits to the possible objects of pro-attitudes, a 
man may have a pro-attitude towards doing something which 
he would otherwise not want to do just because it would be in 
accordance with a rule. The desire to conform to the accepted 
code, to do what others do, to be an inconspicuous member 
of the herd, is an important though not always a decisive factor 
in conduct. And we expect others to be influenced by the 
same motive. If this were not so there could not be such a thing 
as an accepted moral code. 

It is against this background of conformity that advice is 
given and taken. We have seen that it would be improper to 


use even so subjective a vv^ord as 'nice' in an impersonal sen- 
tence except against a background of general agreement ; and 
the same applies, with even more force, to 'right'. Smith asks 
Jones for advice because he believes Jones to be v^'ise and 
experienced in the ways of the world; and if Jones gives 
unequivocal advice. Smith is entitled to infer not only that 
Jones himself approves of the lines of conduct he calls 'right', 
but also that it is generally approved. If there is no such 
general agreement or if Jones knows that his moral view on 
this question is an unusual one, he ought to express it in a 
more guarded way which discloses its peculiarity. Suppose 
that Jones is, unknown to Smith, a conscientious objector, and 
Smith asks whether he ought to join the army. Jones certainly 
ought not to say 'yes', even though he knows that this is in 
accordance with the accepted rule ; for this answer implies that 
he approves. On the other hand a plain negative would equally 
be an abuse of the conventions under which advice is given, 
since it would imply that the consensus of reputable opinion 
is on his side, which he knows to be false. He ought either 
to add 'that is my personal view; but many good people think 
otherwise' or to tell Smith to consult his own conscience, 
v/hich is a refusal to give advice. 

There is one peculiarity which 'right' shares with 'good' 
and 'ought' and with no other words. Except in ironical or 
other secondary uses it is always a pro-word. We saw that 
while A-words must always carry a pro-force when used in 
the contexts of choosing and advising, they do not always 
carry the same force, and we have to gather whether they are 
pro- or con-words from the context. This is never the case 
with 'right'. Some words that are akin to 'right', such as 'just', 
'fair', and 'honourable', are almost always pro-words, but they 
admit of exceptions. They imply a code of rules but they 
could be used in cases where we want to advise a man to 
disobey the code. We might think, for example, that this was 
a case in which justice should give way to mercy and we 
should then say: "It would be the just thing to do; but in this 
case I don't think you ought to do it". But the word 'right' 



is seldom, if ever, used in this way. There is an air of self- 
contradiction about 'That is the right thing to do; but you 
ought not to do it' which is absent from 'that is the just 
(fair, honourable, gentlemanly, etc.) thing to do; but you 
ought not to do it'; for 'right', unlike these other words, is 
always used to give a verdict and not merely to draw attention 
to a reason for giving that verdict. 


'Ought'. Sentences containing the word 'ought' have a com- 
plexity similar to that of A- and G-sentences. They are used 
for a variety of jobs and their contextual implications in each 
case include just those elements that must be implied if they 
are suited to the job they are doing. They cannot be construed 
in a purely subjective or performatory way because, although 
they are used for linguistic performances such as choosing, 
advising, exhorting, and commanding, they are only so used 
under certain fairly rigid conditions. If someone says "I 
happen to like doing X" it makes no sense to ask him why he 
likes it; for the phrase 'happen to' has just the role of indi- 
cating that he can give no reasons. But if he says "I ought to 
do X" he lays himself open to a request for reasons and he 
must be prepared with an answer. It is because we want to 
mark a contrast between a decision based on reasons and a 
decision not so based that we have these different forms of 

As in the case of 'because I thought it best', 'because I 
thought I ought to' does not give a reason for my choice; 
it gives too good a reason. But it indicates that the choice was 
a reasoned and not a casual one and, like ' because I thought 
it right', it indicates the type of reason, namely that it is a 
pro-attitude towards obeying a rule, whether the rule be a 
mere rule of procedure, a rule of etiquette or good manners, 
or a moral rule. 

Similarly, 'You ought to do X' cannot be equated with 



'please do X'; nor is it simply a stronger form of it which 
we use to give weight and authority to our requests. If this 
were the whole story it would be impossible to see how any- 
one could have come to believe that the 'you ought' formula 
is more weighty and authoritative. If someone says "Please 
shut the door" it makes sense to ask for a reason for his 
request. But by way of reply he can get away with " I would 
like it shut" ; if he uses the 'you ought to' formula he must be 
prepared to support it in a very different way. 

Commanding. It is sometimes said that G-sentences and 
'you ought '-sentences are disguised imperatives and, since 
one of the most frequent uses of the imperative mood is to 
issue orders, that they are disguised commands. But this is 
an oversimplification, not unlike the theory that ' this is good ' 
means 'I like this'. The truth contained in the theory is that 
both imperatives and ought-sentences are used for telling 
someone to do something; but they are used in different 
circumstances and the logic of their use reflects these dif- 

A command can (logically) only be issued by someone who 
is competent to issue it, by a person in authority. In practice 
people who are not in a position to use the language of com- 
manding often do so in order to give the impression that they 
have the necessary authority or simply to get something done, 
relying on the well-known docility and habits of obedience of 
their fellow-men. But these are secondary uses, parasitic on 
the proper use of commanding language. If someone issues 
an order it always makes sense to question his right to issue 
it, even if it is in practice impolitic sometimes to do so. But 
commands differ from ought-sentences in that a man who 
gives a command is not logically bound to give any reasons 
why it should be obeyed. On the other hand if a man says 
'You ought . . .' or 'It's worth . . .' he must (logically) be 
able to give reasons. These G-sentences are to commands 
roughly what impersonal pro-sentences are to expressions of 
taste or approval. 

Nor can ought-sentences be identified with commandments, 



that is to say general commands issued to all and sundry and 
to be obeyed not once and for all, as commands are, but 
always. 'Soldiers will salute officers at all times' and 'Thou 
shalt not kill' are orders, edicts, or commandments issuable 
only by a competent authority who is not logically bound to 
give any reason for issuing them or even to have any reasons 
beyond a desire to issue them. But we use 'you ought' sen- 
tences precisely when we are not in a position to issue orders ; 
and this fact and the fact that these sentences must be backed 
by reasons provide an important clue to their logic. Although 
'advice' is far too weak a word for many 'you ought' -sen- 
tences, their logic in primary cases is always that of advising, 
never that of commanding. For they are addressed to a rational 
agent as solutions to his problem of choice and, in consequence, 
they imply a pro-attitude on the part of the recipient. The 
author of a command, on the other hand, is not logically bound 
to be concerned with the pro- and con-attitudes of his subor- 
dinates, though, of course, he may be. 

It is for this reason also that it is a mistake to try to define 
moral 'oughts' in terms of God's commands or 'God's will'. 
For the mere fact that a command has been issued by a com- 
petent authority, even by God, is not a logically good reason 
for obeying it. Jones's 'thou shalt' does not entail Smith's 
'I shall'; and neither does God's 'thou shalt'. We do in 
practice often appeal to the existence of a command in backing 
up an ought-sentence, in the same sort of way that we appeal to 
the existence of a rule ; but this procedure contextually implies 
a general pro-attitude on the part of the recipient towards 
obeying the commands of that authority, as such. He might 
have no other reason for doing X, but decides to do it just 
because he has been told to by such an authority. For religious 
people the fact tha God has commanded them to do some- 
thing is a sufficient i. .;son, perhaps the only reason, for think- 
ing themselves obliged to do it. But this is because they have 
a general pro-attitude to doing whatever God commands. And 
if this were not so, the mere fact that God commands some- 
thing is no more a reason for doing it than the fact that a 


'right' and 'ought' 

cricket coach telis you to do something. No doubt, God, Hke 
the coach, will not tell you to do things that you ought not 
to do, but the fact that you ought to do it cannot be identified 
with nor is it entailed by the fact that God has commanded it. 
The imperative theory of 'ought' fails for the same sort of 
reason that, as we saw in chapter 3, the objective-property 
theory failed. Unlike the objective-property theory it recog- 
nizes the practical role of moral language; but it neglects to 
notice that except in secondary cases, such as exhortation and 
the aping of authority, practical sentences must constitute 
solutions to a problem of choice. This defect is partly due to 
the philosopher's habit of talking about 'obligation' instead of 
talking about 'I ought' and 'You ought'. 


'/ ought' and ' You ought'. Most of the verbs used in ordinary 
descriptive discourse are neutral in respect of pronouns. If 
Jones says truly to Smith "You are sitting on a chair", Smith's 
reply "Yes, I am" and Brown's comment "Yes, he is" must 
also be true. The office of the pronoun is to indicate the person 
to whom the verb applies and the truth- value of the statement 
is not altered by a change of pronoun appropriate to the person 
making the statement. But what is true of the logic of descrip- 
tive discourse is not necessarily true of the logic of practical 

If Jones says to Smith "You ought to do X" and Smith 
replies "No, I ought not", are they contradicting each other? 
Must we say that one is necessarily correct and the other 
mistaken? We have seen that sentences used for registering 
decisions or giving advice contextually imply certain causal 
and predictive elements which are indec d true or false ; so that 
if the dispute between Smith and Jones is found to be con- 
cerned with any of these elements, they are contradicting each 
other and one of them must be mistaken. Among these ele- 
ments in the case of ought-sentences would be the existence 



of a rule recognized by both parties or a command issued by 
an authority whom they both recognize. These are objective 
questions of fact and there is no reason for saying that, if the 
dispute is about them, Smith and Jones are not contradicting 
each other. 

But it is impossible that elements of this kind should be the 
sole elements in the use of ought-sentences. If they were it 
would be impossible to understand the role that these sen- 
tences play in deciding, advising, preaching, and exhorting. 
For the first of these activities it is necessary that Smith 
should have a pro-attitude towards doing what he thinks he 
ought to do ; otherwise ' I ought ' is irrelevant to his problem 
of choice. For the others it is necessary that Jones should have 
a pro-attitude towards Smith's doing what he tells him that 
he ought to do. Otherwise he is involved in the absurdity of 
urging Smith to do something that he (Jones) does not want 
him to do. Human conduct is only intelligible on the assump- 
tion that people intend the natural consequences of their 
actions; and to say 'you ought' is to act, to intervene in the 
world, not to describe it; so that, except in secondary cases 
- for example if Smith is known to be contumacious, counter- 
suggestible, or pig-headed - it is implied that Jones intends 
Smith to do what he tells him. 

Now since Jones's and Smith's attitudes to the proposed 
course of action may differ, it is possible for them to agree on 
all the objective contextual implications and yet for Jones to 
say, rightly, "You ought", while Smith says, rightly, "No, 
I ought not". These remarks can both be logically right in 
the sense that each is the correct way of expressing what the 
speaker intends; whether they can both be morally right is 
another matter. 

It might seem that this situation is still analogous to a 
dispute about an empirical matter. For if Jones says "It's a 
dog" and Smith says "No, it isn't", they can both be right in 
the sense that each has used the correct form of words to ex- 
press what he intended to say, but they cannot both be right in 
the sense of saying what is true. But, as we have seen, there is in 


'right' and 'ought' 

empirical matters a test of truth which is independent of the 
beliefs of a speaker, namely correspondence with the facts, so 
that in these cases there is a point to the distinction between 
the two senses of 'right'. But if the dispute is a moral one 
there is no test of this kind. 

Now it is true that in ordinary life we should say that Smith 
and Jones were contradicting each other. But this only illus- 
trates the danger of drawing philosophical conclusions from 
ordinary language. 'Contradicting', which literally means 
'speaking against', can be used of almost any kind of verbal 
disagreement; but it also has a technical logician's use which 
was designed to elucidate empirical discourse. And if we speak 
of conflicting moral attitudes as 'contradictory' we run the 
risk of unconsciously assimilating moral disputes to em- 
pirical ones and of inventing in the logic of moral discourse 
elements analogous to those which are bound up with the 
notion of contradiction in empirical discourse. One of these 
is 'correspondence with the facts'. And this is to court disaster. 

And disaster has followed. Philosophers have been led to 
talk in terms of a sentence in the form ' X ought to do Y ' which 
is neutral in respect of pronouns and, if true when spoken by 
one man, must also be true when spoken by another. The 
temptation to do this is due to the fact that it does no harm 
when the moral dispute is in fact a disguised empirical dispute 
about the existence of a rule or command which both parties 
agree ought to be obeyed. But we have seen that the use of 
ought-sentences in moral disputes is not confined to such cases. 
And if, as I have suggested, 'Smith ought to do Y' when 
spoken by Smith expresses a decision, but when spoken by 
Jones expresses an injunction, it cannot be extracted from its 
context and pronounced true or false. 

'He ought'. If Jones says "You ought" (giving advice) and 
Smith says "No, I ought not" (deciding to reject it), we 
should naturally say that one of them must be wrong or mis- 
taken or even that what one of them says is false, thus once 
more assimilating the logic of moral discourse to that of 
empirical discourse. But the typical philosopher's phrase 'we 



should naturally say' conceals an important ambiguity. It may 
mean (what it is intended to mean) that 'Smith is wrong' is 
in English a natural and proper formula for anyone to use if 
he wishes to express his moral disagreement with Smith and 
agreement with Jones. But the same phrase could also be used 
to make a moral comment, to express 'our' disagreement with 
Smith'; for example in the sentence ' If Smith said that cruelty 
was right, we should naturally say that he was wrong'. The 
philosopher's point about the use of language must be sharply 
distinguished from the moralist's siding with Jones ; for, while 
the former is true or false in a quite straightfonvard sense - it 
is a matter of historical fact how words are used - the latter 
is not. 

There is this similarity between the moral and the empirical 
case that, since Jones's attitude conflicts with Smith's, Brown 
cannot side with both of them ; and since his logical inability 
to do this is like his logical inability to agree with each of 
two people who contradict each other on a point of fact, we 
mark this similarity by using the same sort of language, 
correct, mistaken, true, false, etc., in the two cases. But this 
is as far as the similarity extends. 

'True' and 'false' are used in moral discourse; and this is 
not an accident. Their use is identical with their use in other 
contexts, in that they are used to endorse what someone has 
said or to endorse its contradictory; and they are not used 
just to endorse, but contextually imply reasons for giving the 
endorsement. We must again distinguish between what a 
man does with a sentence and the conditions which make his 
use of that sentence for that job a proper use. When a man 
says of someone else's empirical statement 'that's true' he is 
not saying that it corresponds with the facts; he is endorsing 
it, siding with it or expressing his agreement with it. But, in 
empirical cases, one of the conditions which make the use of 
'that's true' proper is that he should believe that the other 
man's statement corresponds with the facts. In other contexts 
there are different conditions; for example in mathematics 
there is no question of correspondence with facts ; and in moral 


'right' and 'ought' 

contexts there is no question of this either. In every case the 
conditions under which it is proper to say that what someone 
else said is true are the same as the conditions under which it is 
proper to say that thing. But the conditions are different in 
different types of case. In a moral context it is proper for Brown 
to say of Jones's advice ' that's true ' if it would be proper for 
him to give the same advice, which is what he is, in eflfect, 
doing; but it is not necessary that he should believe that 
Jones's advice corresponds to the facts, because pieces of 
advice are not statements and neither correspond nor fail to 
correspond to facts. 

Just as the logic of discourse about objective properties is 
intelligible only against a background of empirical agreement, 
so the logic of practical discourse is intelligible only against 
a background of agreement on moral principles. It is this 
similarity that leads us to use the same words 'true' and 'false' 
in both cases ; but the background - what has to obtain to make 
the logic of our talk intelligible - is different in each case. 

I shall return to the distinctions between 'I ought' and 
'You ought' when I come to examine Conscience, the opera- 
tions of which seem to combine the two. 



Duty and Obligation 


We have seen that the primary use of 'you ought' sentences 
is to tell someone to do something in cases where it is con- 
textually implied that the speaker has reasons for what he 
says. And we have also seen that the logic of advising, exhort- 
ing, and commanding requires a pro-attitude on the part of 
the speaker to the advice being taken or the order being carried 
out. This need not, of course, be present on all occasions. Any 
use of language can become a mere habit ; a corporal may pass 
on an order from a sergeant without caring a straw whether 
or not it is obeyed, and advice may be given in a similar way 
as a matter of routine. But such uses, however common they 
may be, are secondary uses, not unlike the use of ' good ' when 
a man judges something in a professional capacity without 
himself having any interest in it. 

Since this is so, and since we learn what our duties are by 
being told what we ought to do, it has always seemed natural 
to represent duties as the demands made on us by others. It is 
hardly necessary to say that any attempt to define 'duty' in 
terms of the demands of others is bound to fail, since any such 
definition cannot but fail to do justice to other contextual im- 
plications of the word ; but there is an obvious element of truth 
in this view. Its crudest expression is to be found in the theory, 
which Plato ascribes to Thrasymachus, that my duty is what 
is advantageous to others and, by implication, disadvantageous 
to myself. Now it is clearly false that in all cases in which Jones 
tells Smith to do something, the thing that he tells him to do 
is advantageous to Jones and disadvantageous to Smith. I have 
already cited as examples the case of a man advising another 
on the choice of a career and the case of a father giving 



death-bed instructions to his son. Plato - perhaps deliberately - 
confuses the fact that Jones must have a pro-attitude towards 
Smith's doing what he tells him with the question whether he 
(Jones) is going to reap any advantage from it, and thus gives 
a cynical and egoistic twist to an obvious platitude, namely 
that a man who says 'You ought . , .' has some pro-attitude 
towards the compliance of his audience. It would be logically 
odd if he were sorry to see his advice taken. 

And there is a second contextual implication that is always 
present in cases where someone tells another to do something. 
Although we can and sometimes do use language without hope 
of success, this is a secondary use. In primary cases, Jones 
must believe that there is a considerable chance of his advice 
being taken or his instructions carried out. He must believe, 
that is to say, that Smith will make the required transition 
from ' You ought ' to ' I ought ' ; and this implies that Smith has 
a pro-attitude towards the suggested course. This pro-attitude 
can be of various kinds, (a) In cases of advice, in the usual 
sense of the word and not the wider sense in which I have been 
using it, it was assumed that Jones is trying to help Smith 
solve his problem and that his advice merely takes the form of 
showing Smith how he can implement his existing aim. This 
is what happened in the conversation about the film in Chapter 
II. (b) But we also saw in that chapter that Smith may 
have a pro-attitude towards doing whatever Jones tells him as 
such ; if Jones tells him to do X, he may now want to and decide 
to do X, even though he was previously averse to it or had not 
contemplated it. And this situation is much more typical of 
moral advice ; for we are there concerned not with advice about 
how to achieve an end but with advice about the propriety of 
pursuing the end. 

We must now consider a third type of case. Jones may be in 
a position, not merely to give advice, but to oblige Smith to 
carry it out. Now 'obliging' does not, as we shall see, mean 
compelling. Jones does not have to use force, and if he does, 
then Smith does not choose to do what he is made to do, and 
the case is of no interest to the moral philosopher. But Jones 



may threaten Smith with consequences that he does not want 
or he may point out that, in the society in which Smith hves, 
such consequences are likely to occur. He may back up his 
'You ought' with 'If you don't, I shall . . .' or 'If you don't, 
somebody will . . .'. In these cases Jones provides Smith with 
a motive for doing what he is told. 

The reasons fon which a man does his duty are many and 
various and cannot be profitably discussed at this stage. I am 
far from suggesting that fear of the consequences of dis- 
obedience is the only or even the commonest motive. But it is 
not for nothing that philosophers have spoken of a conflict 
between 'duty' and 'inclination', however absurd some of the 
conclusions they have drawn may be. The language of 'You 
ought' and particularly of 'duty' is frequently used in cases 
where the agent has no reason for doing what he is told other 
than the fact that it is his duty. He has no pro-attitude towards 
the course of action as such. And there is a reason for this. 


The connexion between duties and the demands of others 
comes out clearly in the fact that we use the word 'obligation' 
as a synonym for ' duty ' ; and this word is derived from a root 
meaning 'tied', an obvious metaphor for coercion. My duties 
are what I am obliged to do and it is no accident that inoral 
rules as well as the laws of the land are backed by sanctions. 
The language of moral obligation also contains many words, 
such as 'law', 'must', and 'necessary' which are also used both 
in connexion with logical entailment and causality. Philoso- 
phers have often drawn attention to the analogy between moral 
obligation and logical or causal necessity. Some have tried to 
identify it with one or the other of these; but others have seen 
that this will not do. If it were logically or causally necessary 
that I should do X, then it would be impossible for me not to 
do X and there could be no such thing as moral wrong-doing. 
So they have tried to represent moral obligation as a "third 



kind of necessity". After all, why should there be only two 
kinds ? 

Obligation by circumstances. When we talk about being 
obliged by circumstances to do something we seldom, if ever, 
mean literally 'forced'. We say 'the fact that the road was 
flooded obliged me to make a detour'; but we do not mean 
that we were bodily pushed round the longer route. Obliga- 
tions of this sort prevent our carrying out our original plans 
rather than force us into doing something else. And they 
prevent us in two ways. There is the less interesting case in 
which I just can't continue along the road, because a landslide 
has washed it away, or because it is blocked by some object 
too heavy for me to move. And there is the more interesting 
case in which I could go forward, but the obstacle makes it 
undesirable to do so ; for example the road is unsafe or would 
take, in its present condition, even longer to travel than the 
detour. (Compare: "I was obliged to stop because my car ran 
out of petrol" and "I was obliged to stay a night in Birming- 
ham because I knew I shouldn't reach Oxford before two in 
the morning".) 

The second type of case is the more interesting since in it 
I choose to do what I am ' obliged ' to do, and if it is true that 
ethics is only concerned with what we do voluntarily it is here 
that the analogy with moral obligation must be found. 

There is a sense of ' voluntarily ' in which a man who chooses 
to do something acts voluntarily, even though he is 'obliged' 
to choose as he did. This is the sense in which doing some- 
thing voluntarily and, a fortiori, deciding or choosing to do 
something are opposed to such phrases as 'I can't' or 'I 
couldn't help . . .', when this latter phrase is used to indicate 
physical impossibility. Thus the sailor who is obliged by a 
storm to make for the nearest port, nevertheless decides to 
make for the nearest port; his situation is unlike that of the 
sailor who is simply driven before the wind. Similarly the 
man who is obliged to make a detour because there is a 
flood or a lion in his path decides to go round by the longer 
way. But there is another and much commoner sense of ' freely' 



and 'voluntarily' in which a man who is obliged to do some- 
thing does not act ' freely '. In the first type of case, if you asked 
him why he did not continue on his original route, he might 
reply: "I just couldn't; the path had been washed away". 
In the second type of case he would, if pressed, reply : " Well, I 
suppose I could have gone on; but the R.A.C. man told me 
that the road was very dangerous . . .". And he would add 
that, although he decided to take the longer road, he hadn't 
got zfree choice. The notion of having a free choice or doing 
something as a result of a free choice is much narrower than 
that of choosing to do it. But it is not easy to pin down the 
difference and certain pseudo-problems have been raised by 
ignoring it. 

The sense of 'free' which I shall now consider is one in 
which the word is opposed to 'obligatory' and its cognates. 
There are cases in which we are not free to choose, but are 
obliged to do something; and these two ways of putting it 
amount to the same thing. In addition to the cases I have 
mentioned the following cases may help to indicate the sort of 
choices that are not ' free ' choices, 

(a) I was obliged to move my rook to Q2 to protect 
my queen. (Otherwise my queen would have been 

(b) I was obliged to raise my bid. (Otherwise someone else 
would have bought the house.) 

(c) I was obliged to look it up. (Otherwise I should have 
probably got the answer wrong.) 

There are three things which these cases seem to have in 
common. In the first place I could in any of the cases have 
omitted to do the thing I said I was obliged to do; but, 
secondly, the consequences of doing this would, in each case, 
to put it broadly, have been worse than the chosen course. 
Thirdly, in each case there is a stated or implied reference to 
some special factor that obliged me to take the course I did. 
The phrase ' I was obliged to do X ' has in these cases, not the 



force of ' it was impossible for me to do anything else ' but ' it 
would have been foolish or unreasonable for me to have done 
anything else'. This is a distinction which is easy enough to 
understand, but almost impossible to put in a general form that 
accurately reflects our use of language. And the reason is 
that, although the distinction is clear, it is in practice always 

There is, so far as I know, no phrase used to indicate the 
physical impossibility of doing something that is not also used 
to indicate that doing it, though possible, would be un- 
reasonable or foolish. Thus we even talk about it being 'im- 
possible to get angry with Jones, because he is so kindhearted' 
or 'impossible not to smile'. And the same applies to all the 
quasi-synonyms such as ' I couldn't help ...',' it was out of the 
question ...',' I was forced to . . .', and ' I had to . . .'. And 
there are two reasons for our peculiar habit of allowing 
phrases indicating physical impossibility to be extended to 
cover cases where it is possible, though unreasonable, to do 
something. In the first place we do not always know just what 
we could do if we tried. The man who says ' It was impossible 
not to smile' really does not know whether he could have re- 
frained from smiling if he had tried. Secondly, these phrases 
are often used in order to exculpate oneself from a possible 
criticism ; and since the one cast-iron method of exculpation is 
to show that the suggested alternative was literally impos- 
sible, it is easy to see why people wish to represent cases which 
are really cases of great difficulty or cases where the alterna- 
tive would have been unwise or unreasonable as cases of im- 
possibility. They are trying to justify their conduct. 

The contrast between doing something freely and being 
obliged to do something is made when the circumstance that 
' obliges ' is exceptional and not taken into consideration when 
the original action is planned. Everything that we do is done 
under conditions that are in some way set for us by natural 
circumstances ; even in the most propitious circumstances the 
sailor must take on board enough fuel and set his course 
correctly. It is when the limits set by natural circumstances are 



narrower than usual or narrower than they were expected to 
be that we speak of them as ' obliging ' us. For example we talk 
of being obliged to alter course, but not of being obliged to 
take the most direct route. If I wanted to visit a neighbour 
who lives a mile away, it would be odd to talk about being 
obliged to travel the intervening mile. I am obliged to protect 
my queen when my opponent makes an unexpected attack, but 
not to make a move that is part of my own planned attack. 

Since everything that we do is done under limits set by 
natural circumstances, it is fatally easy to extend the concept 
of obligation by circumstances in such a way as to make it 
useless and to create pseudo-problems. If the brigands on the 
path oblige me to make a detour, might we not say that the 
fact that the normal road is twice as long as the route as the 
crow flies obliges me to do so? We might; but only if we 
wished to draw special attention to the fact that the normal 
road is more devious than most roads are hereabouts. If you 
live in very mountainous country, where all roads are cir- 
cuitous, you expect such things and don't talk about being 
obliged to walk two miles to reach a place a mile from your 
starting point. 

Before comparing cases of being obliged by circumstances 
with cases of moral obligation it is necessary to say a word 
about the ways in which one's own motives -can be said to 
'oblige' one, since the fatal tendency to generalize from the 
exceptional case to all cases has led to some curious philo- 
sophical theories in this connexion. 

Each of the cases of the second type of obligation (but 
obviously not of the first or less interesting type) could have 
been put in terms of being obliged by a motive. Instead of 
saying that the brigands or the storm or an opponent's move 
obliged me, I could say that fear of the brigands, etc., obliged 
me. And just as everything that we do is done under some 
circumstances or other, so everything is done from some 
motive or other. If, then, we can say that fear of the brigands 
obliged me to make a detour, why should we not say that hope 
of gain or a desire to visit a friend obliged me to set out in the 



first place ? There is something queer about talking in this way 
and we are clearly on the verge of a philosopher's paradox, 
that is to say of being forced by gradual steps from saying 
something usual to saying something very unusual. 

Before attempting to deal with this paradox, let us examine 
the paradoxical conclusion. We shall be forced to say that 
everything which we choose to do, since it entails having a 
motive for doing it, entails that we are obliged to do it and 
so did not do it freely. This argument has been frequently used 
to show that there is no such thing as free choice at all; and 
some philosophers have accepted this conclusion. Others, 
accepting the premise that all ordinary motives 'oblige' and 
rejecting the conclusion, have tried to escape the conclusion 
by saying that the sense of duty is a very special motive which, 
unlike all other motives, does not oblige. We act freely, they 
say, only when we are not 'slaves to desire'. I shall try to 
show later that, while the sense of duty is indeed peculiar in 
many ways, its peculiarity does not lie in this and that this 
line of escape involves paradoxes of an even more startling 

The paradox we are here concerned with is startling enough. 
We must say that - leaving aside the sense of duty - whenever 
a man does what he wants to do he is ' obliged ' to do it and so 
does not really act freely. If for example I decide to go to a 
concert rather than a cinema, because I like music more than 
films, I am ' really ' a slave. To what am I a slave ? Why, to my 
superior passion for music of course ! 

Now it may be true that philosophers who talk in this way 
about a man being a slave to his desires - not only in the case 
of drug-addicts, but zvhenever a man does what he wants to do 
— have a genuine point to make ^bout human conduct. But the 
one thing that is certain is that they have chosen a most unfor- 
tunate way of putting their point. For we normally contrast 
'doing somxCthing because I want to' with 'doing something 
because I've got to', 'doing what I like' with 'doing what I 
must', 'being free' with 'being a slave'. And it follows that 
anyone who says that what we normally call freedom is ' really ' 



slavery, though he may intend to say something intelligible and 
even true, is using the word ' freedom ' or the word ' slavery ' or 
both in such unusual ways that he cannot be understood to 
mean what he says. 

It is, however, easy to see that, though there may be some 
point to such remarks - for example to draw attention to the 
fact that people are often in the grip of some ruUng passion 
to a greater extent than they imagine - the theory cannot be 
wholly true. It is logically impossible that all motives should 
oblige, since we talk of being obHged only when our desires 
are thwarted, that is to say when we are obliged to do some- 
thing other than what we wanted or intended to do ; and this 
implies that there was something which we wanted or intended 
to do. Obliging motives presuppose non-obliging motives; 
freedom is the ability to fulfil one's aims, not the state of being 

In order to escape from the paradox that all motives ' oblige ', 
we might for a start try distinguishing motives into two 
types, positive and negative. A negative motive will be one 
which is directed towards an object that the agent would prefer 
to be absent and which, if he were endowed with miraculous 
powers, he would simply wish away. For example the sailor 
prefers a calm sea to a storm and the chess player would have 
been happier if the opponent had not made the move which 
obliged him to protect his queen. (I shall ignore the complica- 
tions introduced by the fact that people enjoy contending with 
difficulties; e.g. a chess player enjoys a game against a for- 
midable opponent in spite of the fact that such an opponent 
' obliges ' him more often than a weaker one.) 

The position is now that a person has some aim, something 
that he wants to do. If the doing of it involves a number of 
steps, which are simply means to the end and which he does 
not want to do for their own sake, it follows that he wants to 
reduce the number and difficulty of the steps; for every un- 
necessary step is a step towards which he has a con-attitude. 
If circumstances oblige him to take steps that he would not 
take under optimum circumstances, his motive for taking them 



might be described as an obliging motive. This account en- 
ables us to do justice to the fact that we do talk about being 
obliged to alter course by fear of the storm, etc., but not 
obliged to go to a concert by a desire to hear music. But unfor- 
tunately this line will not do. For the ingenious philosopher 
who is trying to push us into the paradoxical position might 
argue that people would in fact be better off if they had no 
motives whatsoever, so that all motives would be 'obliging 
motives '. Thus, instead of satisfying my desire to hear music 
by going to a concert, would I not be better off if I had no 
desire to hear music to be satisfied and was not, therefore, 
'obliged' to go to concerts? In this way some philosophers 
have actually recommended as the best life a life altogether 
free from desire. 

The flaw in this argument lies in the traditional equation 
between 'pleasure' and 'satisfaction of desire' and in the 
traditional pattern of all voluntary action as consisting in the 
attempt to remove pain or uneasiness. It is true that in cases 
that do conform to this model it might be better for a man to 
have neither the preceding desire nor the subsequent satisfac- 
tion, neither the hunger nor the food, neither the itch nor the 
scratch. But we have already seen that not all motives conform 
to this pattern. And it is primarily in cases which do not con- 
form to this pattern that it is odd to speak of motives as 
' obliging '. 

Since there are pleasant smells and unpleasant smells a man 
might be better off without a sense of smell ; but if there were 
no unpleasant smells, he would not be. The pleasure of smel- 
ling a rose cannot be thought of as a cancellation of the debit 
balance of desire ; it is a positive credit balance ; so also is the 
taste of good food and the conversation of one's friends ; and 
this enables us to modify our account and come closer to the 
distinction between motives that oblige and motives that do 
not oblige. When we do something for its own sake, i.e. be- 
cause it is just what we want to do, our motive is one that does 
not oblige us to do anything; we are not slaves to our own 
desires or to anything else. On the contrary we are acting 




freely because this is the model par excellence of what ' acting 
freely' means. And the circumstances involved in doing what 
we want do not oblige us either. It is when, in order to do 
what we want, we have to do other things that we do not want 
to do for their own sake that we talk of being obliged. And we 
do this most markedly when the circumstances are not only 
untoward but unusual. 

So long as the necessary steps are not abnormally difficult, 
irksome, or roundabout we think of them as being the natural, 
inevitable way of bringing about our end and do not, there- 
fore, think of them as obstacles to bringing it about. It is for 
this reason that it would be odd to talk about ' being obliged ' 
to take the shortest and easiest route; for when we talk about 
being obliged to do something we always have in mind a less 
irksome or arduous or compUcated alternative that might have 
been possible under other circumstances. If a man talks about 
being obliged to take the shortest and easiest route we are 
puzzled to know what he means. What is he contrasting it 
with? Of course he didn't want to take any route at all; he 
wanted to be at his destination ; he would have been better off 
if he had miraculous powers of reaching his destination with- 
out traversing the intervening distance. The merchant would 
make more profit if New York were nearer to London than it 
is. But it isn't, and nothing that the merchant can do can alter 
this. Hence, in making his plans, he takes this into account. It 
is when we have to deviate from our plans because of circum- 
stances which make our task more difficult, irksome, or com- 
plicated than it would be under the optimum realizable 
conditions that we speak of ' being obliged ' by circumstances 
and of motives which * oblige ' us to revise our plans. 

This discussion has been designed to bring out three points, 
(a) Obligation is not identical with but contrasted with 'free' 
choice, (b) Our own motives do not always oblige us; on the 
contrary there must (logically) be some motives that do not. 
(c) Circumstances which oblige us are untoward and unusual 
obstacles; or, rather, though they may be quite usual, they 
are only thought of as 'obliging' when contrasted with less 



obstacle-ridden courses which are within the bounds of possi- 
bility. We must now compare this analysis of being obliged by 
circumstances with other types of obligation. 


Obligation by threats. Threats come into the picture in verv 
much the same way that natural circumstances do. A man who 
does something because he is threatened does something that 
he would not, but for the threat, have done and something that 
he does not, apart from fear of the consequences, have a motive 
for doing. This need not always be the case; he may have been 
intending to do it anyway; and in that case the threat was 
otiose. But such a case is necessarily unusual. Threats would 
never be used unless the threatener believed that his victim 
would not adopt the suggested course but for the threat. 

The other considerations also apply. A man who is threat- 
ened acts ' voluntarily ' in the sense that he chooses to do what 
he does ; but he does not act ' freely ' ; on the contrary the man 
with the gun obliges him. The threat is an untoward and un- 
usual circumstance which interferes with his free choice. 

Legal and Quasi-legal Obligation. By 'quasi-legal' I mean 
obligation to obey rules or laws of a fairly definite kind other 
than the law of the land and to execute orders other than those 
given by the officers appointed and authorized by the State to 
enforce the law of the land. Examples would be school rules 
or the rules of a club or society. These types of obligation are 
all associated with penalties for breaking the rules and this 
alone is sufficient to make them analogous in some ways 
(though not in others) to obligation by threats and hence to 
obligation by natural circumstances. If I do something because 
I am obliged to do it by a school rule, (a) I choose to do it, 
(b) my choice is not a free one; for I would not have done it 
but for the rule, (c) I have no motive for doing it apart from 
the rule, and (d) I must have some motive for doing it. 

My motive may be fear of the consequences of disobedience ; 

H 209 



but there is no need to be so cynical as to suggest that this is 
the only motive for obeying a rule. I may obey it because I 
have been trained to and obedience is now a matter of habit, 
and in this case I have no motive at all. But I may also obey 
it from a sense of duty; and this is a motive. 

Moral Obligation. There are obvious connexions between 
legal and moral obligations; some moral laws are part of the 
law of the land though they are not made so by statute ; some 
statute laws forbid behaviour that is morally wrong as well as 
criminal, though others do not ; and disobeying the law of the 
land is thought to be morally wrong as such, even if the be- 
haviour which the law makes criminal would not in itself be a 
breach of a moral rule. But, apart from these connexions, there 
are also similarities enshrined in the language that we use in 
common in the two cases. Both (moral) and (legal) laws 
enjoin and forbid; breaches of them are both wrong; and 
breaches are attended with penalties, though the penalties are 
different in the different cases. 

Like other forms of obligation, moral obligation limits the 
range of free choice. I am not free to accept an invitation to 
dinner, because I am obliged by my duty as a member of the 
local Conservative Association to attend a meeting or because 
I am obliged by having made a promise to dine with someone 
else. I am not free to marry the girl of my choice, because I 
know that it would break my mother's heart and I am obliged 
by filial duty. A moral obligation is, like a natural obligation, 
something which obliges me to act in a way that, but for the 
obligation, I would not have acted. It is analogous to the 
second rather than to the first type of natural obligation, since, 
although I have not a free choice, I could always break the 
promise or my mother's heart. In fulfilling a moral obligation 
a man chooses to do what he does, but does not choose freely. 
The feature which distinguishes moral obligations from all 
others is that they are self-imposed ; I shall discuss this 
point in a later chapter. 

In moral, as in the other, cases the logic of obHgation re- 
quires a conflict between the obligation to do something and 



the inclination not to do it. But it is important to notice that 
this conflict is part of the general background of the concept 
of obligation and need not occur in every case. 

It is sometimes said that a man cannot have a duty to obtain 
pleasure for himself- and we must remember that philosophers 
use the word 'pleasure' in a very wide sense. But this is not 
quite true. If we thought that a man was ruining his Ufe by 
excessive abstinence, we might well say that he ought to pursue 
pleasure more than he does. But there is, as usual, some 
truth underlying this view. It is paradoxical to tell someone 
that he ought to pursue pleasure more than he does; but this 
is only because most people do not need telling. 

Codes of law and moral codes do not usually contain injunc- 
tions to do or to refrain from things that people would do or 
refrain from in any case. As we shall see the function of such 
codes is to provide people with motives for doing what they 
would otherwise not do. Thus, although a man may do his 
duty willingly or even gladly in a particular case, what he does 
would never have been incorporated in a code of duties if there 
were not a general presumption that the type of action con- 
cerned is one that people are, on the whole, disinclined 
to do. 

If, for example, people were seldom or never disinclined to 
tell the truth we should never have come to regard telling the 
truth as a duty. In the same way people are, in general, not 
inclined to part with their money and if this were not so we 
should not think of paying a debt as a duty. We frequently 
promise to do things that we are otherwise inclined to do, for 
example to meet a friend. If all promises were of this kind, 
they would still be useful for the purpose of making arrange- 
ments for cooperative activities. But part of the point of mak- 
ing a promise is that a man who makes one binds himself to do 
something which, when the time comes, he may not want to 
do ; and if we only promised to do things that we would do in 
any case because we enjoy doing them, promise-keeping would 
not be thought of as a duty. There is therefore a sense in which 
it must (logically) be the case that people who do their duty 



act unwillingly or against their inclination; but there is an- 
other sense in which they cannot be said to act unwillingly. 
How could they, when it is what they choose to do ? 


Learning the language of duty. We learn the language of duty 
at a very early age and, however different the child's use of 
this language may be from the adult's, there remains a con- 
nexion between the two. The latter grows gradually out of 
the former; and if this were not so it would be a mystery how 
adults come to use this language as they do. It is not necessary 
to be an expert child-psychologist to understand the main out- 
line of the way in which children learn the language of duty. 
It is obvious that a child must learn to understand the use of 
a word by others before he can use it himself. This applies to 
all words; but it applies in a special way to such words as 
* ought ' and ' mustn't '. It is perhaps not logically necessary that 
it should be so, but there are obvious practical reasons for the 
fact that a child comes into contact with ' you mustn't ' before 
he comes into contact with ' I mustn't ' and that he learns the 
language of imperatives in the first instance from having them 
addressed to him by others. Telling himself to do things or 
not to do them is a relatively sophisticated performance. He 
learns to obey before he learns to command. 

Now some desires are instinctive, in the proper sense of the 
word, even though a child may have to learn how to satisfy 
them; and in some cases babies can learn how to satisfy a 
desire without being told to try. A baby or young child tries 
to do what he wants to do and his wanting to do it does not 
seem to him to be commanded or even suggested by anyone 
else. On the other hand there are a host of things that others 
want him to do towards which he has no inclinations whatso- 
ever. His conduct can be influenced in two radically different 



(a) We may attach penalties to a type of action that we 
wish to inhibit so that the child prefers not to do it, even when 
he wants to. And the child learns the language of duty in con- 
nexion with such penalties. The word ' mustn't ' has no magic 
power and the child needs no special insight to learn to under- 
stand it. ' What I mustn't do ' is, for him, simply ' What I get 
punished for ' and it is through this connexion of 'mustn't ' with 
punishment that the word acquires its practical force. No child 
is told that he must do, or obliged under penalty to do, what 
he wants to do already; and for the same reason in each case, 
that he already has a motive for doing it and will do it ' of his 
own free will'. The fact that the commands and prohibitions 
are often issued in the child's own interest is quite irrelevant. 
Children have to be 'obliged' to do what is in their own in- 
terest because they are incapable of knowing that they already 
have a good reason for doing it. 

(b) But it is not to be supposed that, in the twentieth cen- 
tury at least, children learn the use of moral words wholly or 
even mainly in connexion with a system of rewards and punish- 
ments. Most, if not all, small children enjoy the favour and 
affection of their parents, and fear more than anything else any 
signs of this favour being withdrawn. It would be wholly 
misleading to represent parental favour and disfavour as 
rewards and punishments. The child simply notices that his 
parents respond to some modes of behaviour in a way that he 
likes and to others in a way that he dislikes. He learns that what 
adults call 'being good' is identical with what makes Daddy 
smile and behave in a generally propitious way, while what they 
call 'being naughty' is identical with conduct that has the 
reverse effects. There is no other obvious common character- 
istic in all ' good ' behaviour or in all ' naughty ' behaviour. 

These facts have an important bearing on what is meant by 
calling conscientiousness an artificial virtue. One of the most 
important and widespread of the natural pro-attitudes of 
human beings is the desire to conform to the habits of the 
group to which one belongs, to do 'the right thing'. Now, if 
the people around him habitually conform to a certain code of 



conduct, the child is likely to contract the same habits and 
hence to do many things towards which he has no natural in- 
clination and which it would never occur to hirn to do if they 
were not habitually done by others. A child does not always 
have to learn to tell the truth either because he recognizes the 
intrinsic value of truth-telling or because lying is associated 
with punishment or parental disfavour. He does it because he 
lives in a society in which it is habitual ; so that learning to tell 
the truth is almost synonymous with learning to talk. But to 
form a habit of doing" something that is in fact part of the moral 
code is not the same as to form the habit of obeying the moral 
code. The imitative instinct or desire to conform will account 
for the former; but not for the latter. 

Conscientiousness is an 'artificial' virtue in a sense in which 
altruism is not. In fact very young children are often selfish in 
that they display a complete indifference to the pleasures and 
pains of others. A child learns to seek its own pleasure and to 
avoid pain long before it comes to think of others as enjoying 
or suffering at all. But children do not have to be taught to be 
altruistic in the way that they have to be taught both to recog- 
nize and to do their duty. Affection for others, that is to say 
a desire for their welfare, comes naturally to most people as 
soon as they begin to think of others as ' people ' at all. This 
natural benevolence can, of course, be inhibited by bad treat- 
ment; it is stronger in some children than in others, and it is 
absurd to ask if people in general are naturally altruistic or 
selfish. Since natural benevolence is, in most people, too weak 
to make them behave with that degree of altruism that we 
want them to acquire, most children have to be coerced into 
being more altruistic than they naturally are. But with 'con- 
scientiousness' it is altogether different. There is no natural 
conscientiousness in the way that there is natural benevolence. 
Children learn to recognize and do their duty, not because they 
are endowed with a special faculty of recognizing it and a 
special, natural propensity to do it, but because they are arti- 
ficially provided with natural reasons for doing those things 
that at the same time they learn to speak of in the language of 



duty. Undesired consequences - to use a conveniently vague 
phrase - are not only levers by means of which people are 
obliged to do their duty ; they are an essential part of learning 
what 'ought' and other deontological words mean. 



Duty and Purpose 


We have seen that 'you ought '-sentences must be backed by 
reasons and that giving the reason often takes the form of 
appealing to a rule. This is not a peculiarity of moral oughts; 
we can appeal to a rule in support of any injunction or advice 
which is given in terms of 'you ought' or of 'right'. Moral 
rules are indeed a special class of rules and I shall discuss later 
the conditions under which we are prepared to call a rule a 
'moral' one; but there do not seem to be any sharp logical 
differences between the way in which moral rules ,are con- 
nected with ' ought ' and ' right ' and the way in which non-moral 
rules are. If this were so it would be difficult to under- 
stand why we use the same words in the different cases, why 
we talk about the right train to catch, the right way to spell 
a word, the right way to address a duke, and the (morally) 
right thing to do. If moral right and wrong were, as some 
philosophers say, new entities, phenomena, or qualities that we 
descry with a special faculty, it would be impossible to under- 
stand how people learn to use moral words and also to under- 
stand the way in which bad manners shade imperceptibly into 
social maladroitness on the one hand and immorality on the 
other. No one is going to postulate a special non-natural pro- 
perty of dukes in virtue of which we ought to call them Your 
Grace, though the obligation to address them in this way is 
' non-natural ' in the sense of artificial or depending on human 

The dispute between teleologists and deontologists turns 
largely on the status of moral rules and their connexion with 
'ought' and 'right'. As we saw in Part I, intuitionists tend to 
adopt the deontological view while naturalistic philosophers 



tend to be teleologists. In this chapter I shall try to show that 
as usual, both parties are right and to put forward a modified, 
teleological interpretation which allov/s due weight to what is 
true in the deontological theory. On this occasion it is the 
deontologists who are, on the whole, right about the meaning 
of moral words but ignore the contextual backgrounds of their 
use, while the teleologists tend to treat this background as part 
of the meaning. 

Rules and Ends. It is characteristic of teleologists to treat 
all ' oughts ' as hypothetical and all rules as rules for attaining 
a given end. But this raises two difficulties about moral rules. 
Some of these are no doubt only binding because of the good 
consequences that they bring about. In these cases we need no 
direct apprehension of an obligation because we can justify the 
obligation by showing that it is only by fulfilling it that we 
can achieve something else that we ought to achieve. But what 
about this second ' ought ' ? It cannot be the case that all moral 
rules depend on other rules in this way; there must, it seems, 
be some rules - perhaps a very small number or only one - 
which cannot be justified by appeal to superior rules. More- 
over moral rules diflFer from other rules in one very important 
way. They are supposed to be obligatory, not just on this or 
that occasion with this or that end in view, but semper, uhique, 
et ab omnibus. If there are occasions on which a particular moral 
rule ought to be disobeyed these can only be occasions when 
the rule comes into conflict with higher rules and the less 
stringent duty must give way to the more. 

In order to do justice to these facts the deontologist must 
adopt one of two courses. He must either try to show that con- 
flict betv/een rules is, in the long run, only apparent and that 
there is either only one basic moral rule on which all others 
depend or that, however many basic rules there may be, they 
are aU consistent with each other. This theory is known as 
Rationalism and represents a system of moral rules as analo- 
gous to a system of geometry or logic. The alternative is to 
adopt the intuitionist standpoint and to say that, while some 
generalization may be possible, in the last resort we discover 



what we ought to do by estimating the relative stringencies 
of moral claims on each occasion. What we know for certain 
is not that we ought to obey general rules but what our duty 
is in a particular case. 

Teleologists cannot adopt either of these courses ; but they 
still have to account for the universality, or at least the wide 
generaUty of moral rules and for the fact that not all rules can 
be shown to be dependent on other rules. From a laudable 
desire to be synoptic they have often tried to find some large 
general purpose which all men necessarily have and by refer- 
ence to which hypothetical rules of a very general kind can 
be justified. Moral rules are to the achievement of happiness 
what the rules that you will find in a manual on How to play 
Cricket (but not in the Laws of Cricket) are to success at that 
game. They refuse to accept such a rule as ' Thou shalt not kill ' 
as the fiat of God or Society or Conscience and claim that, 
however absurd it may be in practice, it always makes sense 
to ask 'Why shouldn't I?'. Moreover they have insisted that 
such sceptical questions can only be answered by showing that 
the course of action enjoined by the rule is, in the end, advan- 
tageous to the agent. This standpoint comes out clearly in 
Plato's way of making Glaucon and Adeimantus ask Socrates 
to prove that it is to the advantage of the agent to be 'just', 
i.e. to obey a certain moral code. The implication clearly is 
that if Socrates cannot prove this there is really no obligation 
to be just at all. Morality is moonshine; but not quite in the 
way that Thrasymachus thought. For Socrates is represented 
as agreeing with Thrasymachus on the fundamental point that 
the obligation to be just must be justified by reference to the 
good accruing to the agent; he differs only in thinking that 
Thrasymachus takes a narrow and short-sighted view of what 
that good is. 

The same point comes out clearly in a famous passage of 
Butler. " Though virtue or moral rectitude does indeed consist 
in affection to and pursuit of what is right and good, as such; 
yet, when we sit down in a cool hour, we can neither justify to 
ourselves this or any other pursuit, till we are convinced 



that it will be for our happiness, or at least not contrary 
to it."i 

This was the famous mistake of which Prichard accused 
almost all moral philosophers ; and, if it is really a mistake, it 
is not difficult to understand the appeal of intuitionism. For, 
if it is a mistake even to try to justify obligation by reference 
to the purpose of the agent, there seems no alternative but to 
represent it as a unique entity or phenomenon of which we are 
immediately aware. I have already suggested in Chapter lo 
that one of the main reasons for abandoning the teleological 
standpoint was that it appeared to entail the theory that all 
conduct, even the best, is ultimately selfish and that this 
criticism rests on a confusion; and I shall now examine some 
mistakes that teleological writers do in fact make. 

The fundamental mistake is that of confusing three distinct 
sorts of questions, logical, factual, and moral. The logical ques- 
tions are those about the meanings of the words used in moral 
discourse and about the relations between moral concepts. For 
example, " Can 'right' be defined in terms of 'good' or 'fulfil- 
ment of purpose ' or is it an irreducible concept ? ". The factual 
questions are mainly historical, sociological, and psychological. 
For example, " What rules do we actually have ? ", " How did 
we come to adopt just this set of rules?", "What do men, in 
fact, desire, enjoy, find pleasant, etc.?". And the moral ques- 
tions are " What ought I to do ? ", " What rules is it best for me 
to adopt?". I shall examine the ways in which these different 
types of question are related in the next chapter; in this chap- 
ter I shall try to show how confusions between them have 
seriously distorted the answers which teleologists want to give 
to them and that these confusions have led to criticisms which 
are sometimes justified and also led to the belief that duties 
must be independent of purposes and directly apprehended. 

I. Sermon XI (my italics). 




(a) Teleologists, in their desire to construct a single all- 
embracing system of morality, have tried to represent all moral 
rules as dependent for their validity on their tendency to pro- 
mote a single end which they call Pleasure, Happiness, The 
Good Life or, since it is obvious that virtue is not always re- 
warded in this world, Eternal Bliss. But, in so doing, they 
have distorted the logic of moral words and their conclusions 
either turn out to be disguised logical truisms or to be false or 
at least questionable. The more concrete and detailed the pic- 
ture they paint of Happiness or Eternal Life the more obvious 
it becomes that no obligation to try to achieve this follows from 
the description. Primitive conceptions of Eternal Bliss are any- 
thing but vague; and just because they are fairly precise it 
makes sense to ask ' Ought I to try to achieve this state ? '. And 
this gives the deontologist his strongest weapon. Paint the 
picture in as glowing colours as you can; it does not follow 
that you ought to try to bring it about. 

(b) The same confusion has led some teleologists into 
representing empirical falsehoods as logical truisms. Thus 
Hobbes thought that to desire one's own death was a logical 
impossibility, although it is obvious that it occurs. There is no 
logical Hmit to the possible objects of pro-attitudes except the 
logical limit of descriptive discourse, which is self-contradic- 
tion; a point first clearly made by Hume.^ So long as a man 
expresses the object of his desire or enjoyment in a way that 
is self-consistent, he cannot be convicted of any logical error. 
We are, of course, entitled to disbelieve him if he says that 
he would rather be roasted by the fire than warm himself at it; 
but this is only because we know that men are more apt to lie : 
or to misrepresent their own desires than to desire any such 


(c) Again, Gay defines obligation as " the necessity of doing 
or omitting any action in order to be happy . . . and no greater 

I. Treatise, Book II, Part 3, Section iii. 


obligation can be supposed to be laid upon any free agent 
without an express contradiction".^ But this is palpably false 
both as a theory of what people mean by ' obligation ' and as 
a theory about the things that people in fact think themselves 
obliged to do. If Gay is right, the whole human race labours 
under a monstrous illusion. Did Regulus return to Carthage 
in order to be happy ? And, if not, are we to say that he was 
foolish, irrational, or immoral ? To meet this sort of objection 
teleologists sometimes say that Regulus must have supposed 
that he himself would be unhappier in Rome suffering the 
pangs of a bad conscience than he would be undergoing the 
physical tortures that he knew awaited him in Carthage. 

But this will not do. In the first place it is difficult to see 
how the notion of ' conscience ' could have arisen on this hypo- 
thesis or why anyone should have a bad conscience and suffer 
therefrom. And secondly, need we look further for Regulus's 
motive than to say that he may have acted from a sense of 
duty or that he may have acted in order to save his country? 
And the first of these motives cannot be represented either as 
a desire to spare himself the pangs of a bad conscience or as 
a means to this. For it is of the essence of ' pangs of conscience ' 
that they can be allayed only by the knowledge that we have 
acted for the sake of doing our duty and they cannot, there- 
fore, be allayed if we know that we have acted for the sake of 
allaying them, unless we deceive ourselves so grossly as to 
mistake the one motive for the other. And in the same way a 
desire to save one's country cannot be represented as a covert 
desire for one's own happiness or as a means to this. It is a 
contradiction to say that a desire to save one's country is a 
desire for one's own happiness (although the same course of 
action might satisfy both desires) and, since Regulus looked 
forward to certain death, he can hardly have desired the salva- 
tion of his country as a means to his own happiness. 

Gay has, in fact, confused the truism that Regulus must have 

had a pro-attitude towards returning to Carthage with the 

empirical (and, in this case, obviously^ false) assertion that his 

I. Selhy-Bigge: British Moralists, Vol II., p. 273. 



motive was to reap some advantage for himself. The truth 
underlying the teleologist's theory is a logical truth, not an 
empirical assertion about the things that people in fact desire 
or enjoy. It is that Regulus must have had some pro-attitude 
towards returning to Carthage ; otherwise he could not be said 
to have chosen to return. The theory that men can only aim at 
their own happiness is plausible only if ' happiness ' is covertly 
used as a general word covering ' whatever men aim at ' . Since 
this conflicts flagrantly with its normal use it is not surprising 
that some teleologists have slipped into covert egoism. 

Gay is not the only or the best known philosopher to have 
made this mistake ; Mill also seems, at times, to have supposed 
that ' right ' means ' conducive to the greatest happiness of the 
greatest number' and Moore expressly says that "the asser- 
tion ' I am morally bound to perform this action ' is identical 
with the assertion ' This action will produce the greatest pos- 
sible amount of good in the Universe'.", although neither Mill 
nor Moore slips into egoism.^ 

To these attempts to discover a single purpose which justi- 
fies all moral rules or to define obligation in terms of purpose 
deontologists have rightly objected that the special role of sen- 
tences including the words 'ought', 'duty', 'obligation', and 
'right' cannot be construed in this way. I have tried to show 
that the reason which they give for their objection, namely 
that deontological words stand for unique, indefinable, and 
immediately apprehended entities, is mistaken. And it is mis- 
taken because the deontologists themselves represent the issue 
as one of fact, when it is really one of the logic of concepts; 
but on the logical point they are right. 


Pro-words and G-words cannot be defined in terms of each 

other; but to admit this is not to say that there is no logical 

connexion between them or to require us to represent G-sen- 

I. Principia Ethica, p. 147. 



tences as statements of which the role is to describe a special 
world of ' values ' or ' duties ' altogether cut oif from the pro- 
and con-attitudes of ordinary human beings, or connected to 
those attitudes by means of a string of necessary synthetic 

Sentences containing pro-words are used to express and 
defend preferences, choices, and decisions. They vary consider- 
ably, as we have seen, among themselves, some having con- 
textual impUcations which others lack. For example, some 
pro-words imply that the attitude is a relatively stable or a 
relatively widely-shared one, while others do not. But what 
is common to all pro-sentences is that they either express 
choices or give reasons for choice that are logically impec- 
cable. Deontological sentences, on the other hand, are used 
primarily for advising, warning, commanding, exhorting, and 
admonishing and almost always contextually imply the presence 
of rules. Their use also implies the existence of certain pro- 
attitudes without which they could not be used for the purposes 
for which they are used. This is the truth underlying the 
teleologists' attempt to reduce G-words in one way or another 
to pro-words. 

To be brief, pro-words belong primarily to the language of 
' I shall ', G-words to the language of ' You ought ', and deci- 
sions never follow logically from imperatives. The logical gap 
between the two languages which Plato and Butler noticed and 
which led them and others to try to explain obligation in terms 
of purpose is often obscured by the Janus-character of all moral 
words. And this is due to the facts of the world in which we 
live. Since it can be assumed in a great many cases that a man 
has a pro-attitude towards doing what he is told and towards 
conforming to the customary code of his society, ' You ought ' 
slides imperceptibly into ' I ought ' and ' I ought ' into ' I shall ', 
and deontological words acquire a pro-force in addition to 
their deontological force. They can be used to defend or 
explain a choice or even to express a decision. One party 
makes much of the essentially practical role of deontological 
words and emphasizes their connexion with pro-attitudes, 



while the other party emphasizes their connexion with rules 
and standards, a connexion which is much more a matter of 
verifiable fact. 

But in spite of the misleading way in which he sometimes 
puts his case, the teleologist has a valid point to make. And 
this is what we should expect when we consider the long and 
distinguished history of this type of theory; for it is difficult 
to believe that all theories of this type are altogether mis- 
taken. His point might be put as follows : Deontological words 
cannot be defined in terms of, or in any way reduced to, pro- 
words. The ' attitude ' of the speaker - or of anyone else - that 
is to say his likings, enjoyings, desires, aims, and interests, may 
not enter at all into what he means by calling something ' right ' 
or his ' duty ' on a specific occasion ; moreover there are many 
occasions when the only relevant attitude is his desire to do 
his duty as such. Nevertheless deontological words are logi- 
cally posterior to teleological or pro- and con-words in a 
different way. The latter form part of the logical background 
without which deontological words would not be intelligible 
at all; but the reverse is not the case. We could imagine a 
world in which people used such words as good, desire, aim, 
purpose, choose, happiness, and enjoy, but in which they had 
no conception whatever of duty, obligation, right, and ought. 
In this strange world people's pro- and con-attitudes would 
be very different from what they are in our world ; they would 
enjoy and desire very different things; but their use of pro- 
and con-language would be recognizably similar to our use. 

We could also imagine a world in which people used pro- 
words and also used the words ' right ' and ' ought ' in a purely 
hypothetical way ; for they might discover that they could only 
achieve their ends by adopting certain courses which they 
would call ' the right course ' or ' the course we ought to adopt '. 
But it is impossible to imagine a world in which people used 
the words obligation, duty, right, and ought but did not use 
any pro-words at all. 

The reason for this is that all these words, both pro-, con-, 
and deontological, belong to practical discourse; and they 



could not be used in the way that we in fact use them in a 
world in which people did not know what it is to choose; 
and this they could not know if they were indifferent to every- 
thing in the universe. Deontological words belong to the lan- 
guage of advising, exhorting, and commanding rather than to 
that of choosing ; but pro-words still form part of their logical 
background. For we should have no use for the language of 
advising, exhorting, and commanding if we were indifferent to 
everything that everybody (including ourselves) did. And we 
should have no use for this language because we should not 
know what it was to advise, exhort, or command. These are 
things that we do and unless we had motives we could not do 

This account of the relation between pro-words and deonto- 
logical words has nothing whatever to say about what men's 
actual attitudes are, still less about what they ought to be. It 
is a logical thesis, not a psychological one, still less a recom- 
mendation to mankind to adopt certain aims or policies. 



The Purpose of Moral Rules 


In the last chapter I suggested that teleological writers have 
sometimes confused logical, factual, and moral questions. In 
this chapter I shall consider three factual questions : 

(a) Why does a man obey a rule on a particular occasion ? 

(b) Why do we have any moral rules ? 

(c) Why do we have the rules we do ? 

(a) Why does a man obey a rule on a particular occasion? There 
are two kinds of explanation which we can dismiss from the 
start. In the first place he may do something that is in fact in 
accordance with a moral rule from a motive that has nothing to 
do with the rules at all ; for example it may be that it is what he 
wants to do. This does not count as ' obeying the rule ' for our 
purposes. Secondly he may obey from force of habit; either 
he cannot break the habit or it does not occur to him not to 
obey the rule. This case can be dismissed on the grounds that 
he has no motive for obedience. When the explanation is given 
in terms of motives, the motive may be of any of the following 

(i) He may treat the rule as a 'hypothetical imperative' 
laying down the best, simplest, most convenient, etc., way of 
achieving whatever it is that he wants to achieve. It is for this 
reason that people obey, follow, or apply the instructions in 
cookery books, manuals on how to play golf, and so on. It is 
clear that the value of such rules depends solely on the empiri- 
cal question whether obedience to them tends to promote 
success ; so that none of our three questions would be difficult 
to answer if all rules were of this type. 

(ii) He may know that the rule has a sanction and be afraid 



of the consequences of breaking it. Tliis would give a complete 
explanation of why a man obeyed a rule on a particular occa- 
sion; but it cannot answer our second or third questions, un- 
less we suppose that a direct desire to make others afraid is 
much more widespread than it would seem to be. Philosophers 
have said that the aim of the law is terror, but not terror for 
its own sake. 

(iii) He may have a desire to conform to the code in use 
in his society, to do the done thing. This is an exceedingly 
common motive and accounts for a great part of our obedience 
to rules of good manners and minor moral rules when this is 
not habitual. But, again, it cannot answer our second question, 
since this motive presupposes a system of accepted rules. 

(iv) He may obey the rule from a desire to obey the rule as 
such, and when the rule is a moral one this motive is called 
the Sense of Duty. This phrase does not refer to a special 
faculty by means of which we learn what our duties are; that 
we learn in the same way that we learn everything else; it 
refers to a special motive that is so important that I shall dis- 
cuss it in full in the next chapter. 

It should be noticed that these possible explanations are not 
mutually exclusive; they may be mixed, and, what is more 
important, the last motive may be mixed with a direct pro- 
attitude towards doing the action enjoined by the rule without 
regard to its being a rule. A man may give money to charity 
both because he wants to and also because he regards it as his 
duty and wants to do his duty. 

(b) Why do we have any rules at all? To suggest a way of 
life in which there are no rules is to suggest something of 
which no one has ever had any experience and hence to indulge 
in speculations of a desert-island type.^ There are indeed cases 
where we can contrast relatively arbitrary action - and by this 
is meant, not deciding by tossing a coin, but deciding each 
issue ad hoc without reference to general rules - with action 
which is bound by rules. Laws may be administered, for 
I. P. 239, below. 



example, either by an arbitrary despot who decides each case 
as he thinks fit, or by an administrator who is bound by rules 
to decide in a particular way. The rival merits of a society 
governed by an arbitrary ruler, who is assumed to be en- 
lightened and benevolent, and one governed by a ruler who 
is not above the law have been debated for centuries; indeed 
this topic was one of the most important topics of political 
philosophy at least from Plato to Hobbes. All actual known 
systems involve a compromise between the two extremes. In 
our own legal system judges are bound by laws and precedents 
and by complicated, traditional rules of procedure which deter- 
mine to a large extent how laws are to be interpreted and 
what constitutes a binding precedent. 

It is hardly necessary to expatiate on the advantages of 
having some rules. Many of the things that we want to do 
involve large-scale operations extending over long periods of 
time. A man who wants to learn to play a Beethoven Sonata or 
to make a fortune or to convert a whole people to a new religion 
is hardly likely to be successful unless he coordinates his 
activities to suit his ends. This is an empirical fact, a feature of 
the world in which we happen to live ; but it is such a prominent 
feature that it is difficult to imagine a world in which this is 
not the case, a world in which large-scale aims could be 
achieved in a haphazard way. 

The harmonious life. Moral rules are necessary for two main 
reasons. In the first place every man has a great variety of 
aims which cannot all be fully achieved because they conflict 
with each other. If a man wants to be prime minister or a 
great pianist, for example, there are many other things that 
he also wants to do but cannot do if he is to achieve this par- 
ticular aim. If this were the only reason for having moral rules 
we could treat all moral rules as rules of success, rules for co- 
ordinating a man's activities in such a way that he succeeds in 
living the type of life that he most wants to lead; for it is 
again an empirical fact that he is more likely to succeed if he 
sets about achieving this type of life in a regular rather than 
a haphazard way. 



This reason for having moral rules is so obvious and so 
important that some philosophers have thought it possible to 
prove that v^^e ought to follow what Butler calls ' cool self-love ' 
rather than 'particular passions'. But this must be a mistake; 
for to say this is to make a moral judgement, to side with the 
* calm ' against the ' violent ' passions and to recommend people 
to follov/ their long-term interests rather than do what they 
happen to want to do at the moment. And these decisions and 
injunctions cannot, as Hume noticed, follow from the empirical 
statement that if a man is to pursue his long-term interests 
satisfactorily he must curb his passions. 

But, although Butler was wrong in thinking that he could 
prove the moral superiority of cool self-love, he was right in 
a way. The language of 'ought', in so far as it is necessarily 
and intimately connected with rules, is appropriate to the 
achievement of long-term interests and not to the satisfaction 
of desires. There are rules, sometimes only very rough rules 
or 'maxims', which must be observed by anyone who wants 
to achieve a long-term end, such as learning to play the piano ; 
but there are no rules for acting on impulse. So that a man who 
asks himself what he ought to do has already decided against 
a general policy of acting on the spur of the moment. 

Social Harmony. But the achievement of coordination be- 
tween a man's own aims is clearly an unimportant reason for 
having moral rules when compared with the need for coordi- 
nating the aims of different people. Indeed, until we mention 
this, we hardly seem to have touched on moral rules at all ; for, 
although we do sometimes talk about duties to ourselves, most 
of our duties are duties to others. There are two main reasons 
for having social rules, (a) To enable people to cooperate 
successfully in activities which, either logically or in practice, 
they could not carry on individually; for example commercial 
enterprises, amateur dramatics^ games, and warfare. There 
must be honour even among thieves if robbery is to pay. 
(b) The aims of different people conflict, and if there were no 
rules for settling disputes the resulting anarchy would be such 
that no one would achieve his aims. Property-rules are rules 



which exist for this purpose, as is shown by the fact that in the 
case of goods which are so abundant that there is no compe- 
tition for them we have no property-rules. There are rules for 
the distribution of water in the desert and where, as in cities, 
there are costs of distribution, but not elsewhere. 

Now property-rules are not moral rules; but they are logi- 
cally prior to many moral rules. There could be no moral obli- 
gation to pay debts if there were no property-rules, because 
there could be no debts. This is not to say that my obligation 
to pay a debt rests on the scarcity of the object concerned. The 
creditor may be in no need of it or easily able to obtain it else- 
where; and in neither case would my obligation be cancelled. 
But unless there was competition for goods in other cases we 
should never have come to include the obligation to pay debts 
in our moral code. 

(c) Why do zve have the rules zve do? We may divide the rules 
actually found in any society into two classes, superior rules 
and subordinate rules. Subordinate rules are those that nobody 
would think of calling absolute or ultimate rules of morality 
or categorical imperatives. The only reason for adopting such 
rules is that they are connected to some superior rule in one of 
two ways. Either they are special cases which follow logically 
from the superior rule in the way that the obligation not to 
make a false income-tax return follows from the general obli- 
gation to tell the truth; or they are supposed to tend to pro- 
mote some very general object, such as the happiness of others, 
that we think we ought to try to promote. In the latter case 
the only good reason for adopting a new rule or adhering to 
an old one is the empirical fact that obeying it tends to pro- 
mote the general object enjoined by the superior rule. To this 
class belong all the rules of etiquette and good manners and 
a great many moral rules, for example those governing sexual 

Superior rules can be subdivided into two classes: (i) the 
general obligation to promote the welfare of others and 
(ii) rules of ' special obligation ', as they have been called, the 



duty to pay debts, keep promises, tell the truth, and to distri- 
bute good and evil according to merit. I shall call the first class 
' duties of beneficence ' and the second ' duties of justice ' ; but 
it must be remembered that ' duty of beneficence ' means an 
obligation to do good to others, not an obligation to feel bene- 
volent or an obhgation to act from the motive of benevolence. 
Acting from benevolence and acting from a sense of duty are 
quite distinct, even though the two motives may be present 

(i) Duties of beneficence. There are two reasons why bene- 
ficence should be considered a duty. In the first place benevo- 
lence is one of our natural pro-attitudes. It is one that 
conflicts with other pro-attitudes and it is one that tends to be 
stronger in our calmer and more reflective moments. If to do 
good to others is one of a man's dominant aims he has a good 
reason for making this type of conduct a duty; for if he does 
so his desire to do good to others will now be backed up by 
his desire to do his duty, which is an exceedingly powerful 
motive. In this way he is more likely to try to do good to 
others even at moments when he does not much want to do 
so, and so come to fulfil his dominant aim more completely 
than he would if he did not adopt this rule. 

But, even if the desire to do good to others were in fact much 
weaker in most men than it is, every man would always have 
a powerful indirect motive for it. He cannot fulfil any of his 
aims without the cooperation of others and people are un- 
likely to do what he wants if he does not do what they want. 
Security from interference and cooperation would be very un- 
stable if we could rely on them only in cases where the parties 
have mutual interests ; they can only be achieved among men 
as we knoAV them if men can be brought to adopt and maintain 
a general system of doing good to others even when they do 
not particularly want to do so. If it be objected that this is to 
give a cynical account of the reasons for doing good to others, 
the answer is that I am not giving reasons for doing good to 
others but reasons for regarding this as a duty. In a world in 
which a regard for the welfare of others was more wide- 



spread and stronger than it is, beneficence would not be a 

(ii) Duties of Justice. It is notorious that utilitarian theories 
have great difficulty in explaining the duties of special obliga- 
tion. For example, they have suggested that what we ought 
to do on a given occasion is the action which will bring about 
the greatest balance of good over evil that we can in the cir- 
cumstances. Now, if this theory is put forward as representing 
the ordinary man's reasons for calling something right, just, 
or what he ought to do, it is patently false. Utilitarians have 
in fact tended to confuse five different, but connected ques- 
tions : '' 

(a) Why does a man in fact obey a moral rule ? 

{h) What sort of reasons do men usually give for obeying a 

rule ? or : Why do they think they ought to obey a rule on 

a particular occasion ? 
(c) Why should a man obey a rule on a particular occasion ? 
{d) Why do men in fact have the moral rules they do ? 
(e) What moral rules ought we to have? 

Considered as an ansv/er to any of the first three questions 
the utilitarian theory is obviously false. We need distinguish 
the first from the second question only in order to allow for the 
fact that men's motives are not always what they think they 
are; and the first question is not of great interest to ethics. 
The second question can only be answered by examining the 
ways in which people do actually defend and justify their deci- 
sion to obey a moral rule. Sir David Ross puts this admirably 
as follows: 

"When a plain man fulfils a promise because he thinks he 
ought to do so, it seems clear that he does so with no thought 
of its total consequences, still less with any opinion that these 
are likely to be the best possible. He thinks in fact much more 
of the past than of the future. What makes him think it right 
to act in a certain way is the fact that he has promised to do 
so, - that and, usually, nothing more. That his act will pro- 



duce the best consequences is not his reason for calling it 

Ross advances this argument to refute the theory that ' right ' 
means 'productive of the greatest possible good' ; but it is no 
less fatal to the theory that the belief that something will pro- 
duce the greatest possible good is our only reason for believ- 
ing that it is right. It is just untrue that most people think that 
they ought to give Christmas presents which are likely to 
cause a great deal of happiness rather than use the money to 
pay a debt to a rich creditor who does not need it, or to a 
drunken one who is likely to do harm with it. And it is also 
untrue that if a man is asked why he thinks he ought to do 
something he will always reply: "because it will produce the 
greatest possible amount of good ". 

Now the third question is not a question of fact, but a moral 
question, and cannot be so easily decided. An extreme utili- 
tarian who says that the belief that an action will produce the 
greatest possible balance of good over evil is the only morally 
good reason for doing it is not necessarily trying to explain 
what 'right' means or telling us what most people think 
right; he may be advocating a special moral outlook. He is 
recommending people to consider only the good and evil 
consequences of alternative actions when making up their 
minds what to do. He may do this in one of two v^^ays. (a) This 
may be the only moral principle that he espouses and is willing 
to recommend to others, or (b) he may admit the desirability 
of considering other points but say that it is desirable to con- 
sider these points only because we are naore likely to achieve 
our ultimate purpose, which is to bring about the greatest 
balance of good over evil, if we do so. It is this last course that 
most utilitarians have in fact adopted ; but they have confused 
[justifying obedience to a rule in a particular case with justifying 
the rule, question (c) with question (e). 

It is as an answer to questions {d) and (e) that utilitarian- 
ism can best be understood. As an answer to {d) it is, as it 
stands, untrue. For custom and tradition and the power of 
I. The Right and the Good, p. 17. 



interested persons account for a great many of the actual moral 
rules of all societies. We have the rules we do, not because 
we have deliberately adopted them as the best set of rules or 
because they were imposed on our ancestors by Philosopher- 
Kings, but because they are traditional. To understand the 
utilitarian theory we must distinguish between causes and 
reasons. Custom and tradition can explain why old rules are 
retained, and their force is made more powerful by the Janus- 
character of moral words. Since the words 'just', 'right', 'de- 
serve ', and ' ought ' are used both to say what the rules are and 
also to defend adherence to them and recommend others to 
adhere to them, there is always an air of self-contradiction 
about any proposal to change the rules; for the conduct laid 
down in the new rule is necessarily unjust or wrong simply 
by virtue of the fact that it contravenes the old rule. Moral 
reformers have often found this a powerful obstacle. 

But custom and tradition cannot account for changes in the 
rules and they do not even begin to answer the question ' For 
what reasons do we have the rules we do?'. To answer this 
question it is necessary to discover the pro-attitude which 
sufficiently explains why people adopt, adhere to, or change 
a rule. In many cases we need look no further than to the fact 
that a sufficiently powerful or influential set of people have a 
pro-attitude towards the inclusion of a particular rule in the 
moral code of their society. Their pro-attitude is a direct one; 
they want the rule adopted. And their power enables them to 
provide an indirect pro-attitude to their subjects in the form 
of penalties for disobedience. This is not to say that they make 
and enforce those rules which are in their own interest, but 
simply that they make and enforce those rules that they zuish 
to enforce. Logically, the ruling class might be wholly altruis- 
tic and enforce only those rules which benefit its subjects. The 
rules which parents enforce on their children provide examples 
of rules which are enforced because a ' ruling class ' wants to 
enforce them but which are adapted to the benefit of the subject 
class. But in practice it is to be feared that the cynical theory 
that laws are made in the interests of rulers is not far wrong. 



Anthropologists are nowadays suspicious of attempts to 
explain moral rules in terms of their value to society. It is, for 
example, impossible to explain the ancient Hebrew taboo on 
the eating of pork as due to the unwholesomeness of the Pales- 
tinian pig. But the mistake of the older anthropologists was 
that of assuming that the ideas of all societies as to what con- 
stitutes the interest of society must have been the same as our 
own; they were not mistaken in thinking that rules are pro- 
mulgated and enforced because they are believed to be in the 
interests of society, or of some class. 

To illustrate this let us consider the various explanations 
that have been suggested for incest rules. Almost every known 
society has some incest rules, although the rules vary greatly 
between societies. Now it is often supposed that these rules 
are adopted because of their supposed biological utility. The 
offspring of incestuous unions will be insane or deficient in 
some other way. But it is impossible that all these rules should 
be biologically useful, since some of them contradict others; 
and a belief in biological utility could hardly account for the 
fact that in many societies a man is forbidden to marry his first 
cousin on one side and obliged to marry his first cousin on the 
other side. Moreover the question at issue is not whether the 
rules are biologically useful but whether they are adopted 
because they are believed to be. People who have not been 
influenced by our own tradition never in fact allege biological 
utility as their reason; without exception they explain and 
defend the rules either by reference to ancient tradition or by 
reference to the belief that social calamities, floods, crop- 
failures, and the barrenness of women will follow a breach of 
the rules. ^ 

And these explanations are consistent with the utilitarian 

I. The origins of and the justifications given for various incest rules 
tiave been examined by Lord Raglan in jfocasta's Crime. Lord Raglan 
shows conclusively that the explanations suggested by sophisticated 
people from Aristotle to Freud neither explain nor justify the rules and 
also that the belief that breaches cause social calamities is ancient and 
almost universal. His attempt to account for the origin of this strange 
belief is* less satisfactory. 


hypothesis. For the fact that the breach of a rule is not followed 
by the dire consequences supposed is not an argument against 
the theory that the rule was adopted because breaches were 
thought to entail these coftsequences. Dancing Sellenger's 
Round does not in fact help to keep the sun on its course ; but 
it is almost certain that the practice originated because it was 
believed to do so. Why this strange mistake should have been 
made is another matter. If we allow for the part played by cus- 
tom and tradition, the utilitarian theory is a plausible answer 
to the question ' For what reasons have men adopted the par- 
ticular rules they have ? '. It is at least a much more plausible 
answer to this question (d) than to the question (b) 'Why 
do men think that they ought to obey a rule on a particular 
occasion?'. We have the rules we do either because we have 
inherited them and stick to them uncritically or because we 
believe them to promote our ends. 


The most important confusion and the one that has given rise 
to most of the misunderstandings of the utilitarian position is 
that between question (c) and question (e). To understand 
the utilitarian position it is necessary to distinguish between 
the judge's question and the legislator's question. Both the 
questions are moral questions, questions about what someone 
ought to do ; but the logic of the answers is very different. 

The duty of the judge is to pronounce verdict and sentence 
in accordance with the law; and the question 'What verdict 
and sentence ought he to pronounce?' turns solely on the 
question ' What verdict and sentence are laid down in the law 
for this crime ? '. As judge, he is not concerned with the con- 
sequences, beneficial or harmful, of what he pronounces. Simi- 
larly, the question 'Was that a just sentence?' is one that 
cannot be settled by reference to its consequences, but solely 
by reference to the law. The logic of the phrases 'just verdict', 
'just sentence', and 'just punishment' requires a reference to 



aws or rules at two points : (a) there must be a law forbidding 
he deed done, and (b) there must be a law attaching a certain 
)enalty to breach of the first law. 

In assessing the justice of a punishment in a particular case, 
herefore, we are concerned only with these two points. The 
junishment is just if and only if the accused committed the 
:rime and the punishment is that laid down by the law for that 
;rime. The question of the probable effects on the accused 
)r on others (the reform and deterrence so greatly empha- 
iized by utilitarians) is relevant only in so far as judges are 
lUowed considerable latitude within which they may be guided 
)y considerations other than what the law prescribes. In prac- 
ice, both in civil and in criminal cases, judges often express 
■egret at being in duty bound to give the verdicts they do and 
ecommend the alteration of the law. 

But the duty of the legislator is quite different. It is not to 
lecide whether a particular application of the law is just or 
lot, but to decide what laws ought to be adopted and what 
)enalties are to be laid down for the breach of each law. And 
hese questions cannot be decided in the way that the judge 
lecides what verdict and sentence to pronounce. For if we 
nterpret the legislator's question as one to be settled by ask- 
ng 'What does the law lay down for such a case?' we shall 
;ither Be involved in an infinite regress, a hierarchy of laws 
n which the justice is determined by reference to a higher law, 
)r we shall be forced to claim intuitive insight into a system 
)f axiomatic laws, themselves requiring no justification, but 
)roviding the justification of all lower laws. 

This last course (which is the classical theory of Natural 
L^aw) is involved in all the difficulties raised earlier in con- 
lexion with intuitionist theories. It is equivalent to saying 
hat we are quite certain that some laws ought to be adopted 
md others not, but cannot explain why. And it cannot explain 
he Janus-character of such words as 'just', 'deserve', 'blame- 
vorthy', etc. For these words have a gerundive force; to say 
hat a law is 'just' is to recommend people to adopt or adhere 
o it ; and if it were merely to state that it had a non-natural 



property, we should still have on our hands the question 
* Why should we adopt this law ? '. 

Now since there are no logical limits to the possible objects 
of pro-attitudes it cannot be proved that men have not got a 
direct and natural pro-attitude towards certain forms of con- 
duct such as paying debts, keeping promises, and punishing 
malefactors. I shall give reasons in the next chapter for sup- 
posing that, although the pro-attitude among people who have 
been brought up to obey certain rules is a direct one, it is not 
a natural one. The belief that it is natural seems to be due to 
the failure of the utilitarian account of what makes a particular 
action just or right. And this in turn is due to the utilitarian's 
confusion of the question ' Why should a man obey a rule on 
a particular occasion?' with the question 'What rules ought 
we to have ? '. 

It is worth noticing that Hobbes, who was a forerunner of 
the utilitarians, was alive to the distinction between the judge's 
duty and the legislator's duty, the difference between the way 
in which a verdict can be appraised as just or unjust and the 
way in which a law can be appraised. But he drew attention 
to the distinction in a misleading way which has led his critics 
to make his theory out to be far more shocking than it really 
is. Laws, he thought, could be good or bad, but not just or 
unjust; for, he said, 'just' means 'in accordance with the law'. 
His way of putting it illustrates the danger of monkeying with 
the usual meanings of words. We do in fact use 'just ' and ' un- 
just' of both verdicts and laws; and although the logic of justi- 
fication is different in the two cases, there is, as always, a good 
reason for our using the same word. We call decisions made 
in accordance with laws just and also call the laws themselves 
just. The reason for our doing this is not unlike our reason 
for calling empirical, mathematical, and moral propositions 
true. 'Just' is a pro-word which indicates our endorsement 
of the decision or of the law ; but our reasons for endorsing it 
are different in each case. The tendency to confuse the two 
cases is partly due to the fact that we often want to do both 
at once. If a man calls a decision just without adding "but, 



nind you, I think the law is a very unjust one", he is ahvays 
issumed to be in favour of the law as well as of the decision. 

To understand utilitarianism we must, therefore, distin- 
guish questions about the reasons for adopting, retaining, or 
iiscarding a rule from questions about our obligation to obey 
;he rule. The obligation to obey a rule does not, in the opinion 
Df most ordinary men, rest on the beneficial consequences of 
abeying it in a particular case in either the short or the long 
run, as utilitarians have almost always supposed. But the 
reasons for adopting a rule may well be of the kind that 
itilitarians suggest. It is, of course, impossible to prove that 
:hey are the only good reasons, since this would be a moral 
judgement. But it is a moral judgement that most men would 
36 much more likely to endorse if it were not confused with 
:he different moral judgement that we ought only to obey moral 
rules when the consequences of obedience are likely to be good. 

It is only fair to the utilitarians to add that they were always 
more interested in the legislator's than in the judge's prob- 
lem ; and this explains their inadequate treatment of the latter. 
Their opponents, on the other hand, have tended to ignore the 
legislator's problem altogether. The fact that the moral rules 
idopted in a given society are believed to promote the ends 
3f the members of that society and the fact that rules tend to 
"all into disuse when the beliefs are discarded are facts which 
Dught to have puzzled deontologists more than they have.^ 


Desert Islands. I have suggested that it always makes sense to 
isk 'Why obey this rule?' even though it is often impolitic 

I. As an example, when we discover that people who behave in a 
certain way are not in fact reformed by punishment, we tend to remove 
;he type of conduct concerned from the list of crimes (although the 
aw, for very good reasons, lags behind public opinion in this respect). 
But the question whether a certain type of conduct ought to be re- 
garded as a crime is of a quite different sort from the question whether 
i man who has done what, as the law stands, is criminal ought to be 
punished for it. (Cf. pp. 272,, 306.) 



to do SO. One of the commonest arguments against this view 
consists in inventing peculiar cases in which it is said to be 
obvious both that a rule ought to be obeyed and also that no 
advantage is going to accrue to anyone from obeying it. This is 
held to show that obedience to a rule neither can be nor re- 
quires to be justified by an appeal to consequences. Suppose, 
for example, that a man makes a promise to another man who 
is dying on a desert island to dispose of his goods in a certain 
way if he ever reaches home. Is he under any obligation to 
keep his promise if it is clear to him that some other distribu- 
tion of goods is going to be more generally beneficial and if 
there-is no chance of his breach of trust being detected ? 

We are invited to believe that the answer to this question 
is obviously ' yes ' ; and utilitarians have tried to reconcile this 
obvious answer with their theory in various ingenious ways. 
They point, for example, to the impossibility of being quite 
sure that the breach of trust will not be detected or that it will 
do more good than harm. They point also to the bad effects 
which such a breach is likely to have on the agent's own 
character. But it is clear that, as long as they admit the con- 
clusion, these attempts at evasion are useless; for the relent- 
less desert-islander will break them down one by one by 
adding stipulations to the terms of the original problem. 

I shall adopt the more radical course of challenging the 
conclusion. It is always difficult to assess the force of these 
desert-island arguments that depend expressly on the im- 
probability of the case supposed, precisely because the case 
is improbable and therefore not catered for in our ordinary 
language. Compare the question ' What would you say if half 
of the standard tests for deciding whether a piece of copper 
wire is electrified gave a positive answer and half a nega- 
tive ? ' ; or the question ' What would you say if you added a 
column of ten figures a hundred times and got one answer 
fifty times and another the other fifty times?'. The answers 
to these questions could only be either " I must see a doctor 
at once " or " I simply do not know what I should say ; for the 
logic of my language for talking about electricity or adding 

240 \ 


does not allow for this sort of thing. If it occurred, I should 
have to treat some sentence which normally expresses an 
analytic proposition as expressing a synthetic one; but I cer- 
tainly cannot say which." 

In the same way I confess to being quite unable to decide 
now what I should say if a desert-island situation arose. Moral 
language is used against a background in which it is almost 
always true that a breach of trust will, either directly or in the 
more roundabout ways which utilitarians suggest, do more 
harm than good; and if this background is expressly removed 
my ordinary moral language breaks down. For it is the back- 
ground which gives the air of self-evidence to the assertion that 
the rule ought to be obeyed. This self-evidence is due to the 
Janus-character of moral words, and the fact that they have 
this character is in its turn due to the normal background of 
their use. 

Suppose it is said that to break the promise would be unjust, 
wrong, wicked, dishonest, or dishonourable. The minimum 
force of these expressions would be to point out the obvious 
fact that simply qua breaking a promise the action falls into 
a class of things to which these epithets are commonly applied. 
But this, though true, is not to the point. The objector clearly 
means to say more than this ; he means to say at least that the 
customary moral code forbids breach of promise even in the 
circumstances alleged. And since customary moral codes are 
not designed to cover cases that are ex hypothesi unusual, this 
is at least doubtful. But the objector probably means more 
than this; he is suggesting that everyone would decide that 
he ought to follow the rule even in the peculiar case, whether 
Dr not customary morality enjoins it. But such a decision could 
never be self-evident unless it was in the trivial form ' I ought 
to do what I ought to do'. Now, since the very words 'just', 
right', 'honourable', and 'dishonourable' have a gerundive 
force as well as a fact-stating force, ' I ought to do the honour- 
ible thing' and 'To keep the promise is the honourable thing' 
30th seem to be tautologous and the conclusion ' I ought to 
ceep the promise ' seems self-evident. But the trick is manifest. 
I 241 


The first premise is only tautological if ' honourable ' is taken 
to be a G-word, the second only if it is taken as a fact-stating 
word about the customary code. Clearly I might decide that 
in this case I ought to do the ' dishonourable ' thing. 

The prototype of all desert-island arguments occurs in the 
second book of Plato's Republic, where Plato suggests that if 
a man had a magic ring which enabled him to escape detection 
and so evade punishment he would be under no obUgation to 
be just. Now it is obvious that, in the world as we know it, 
the obligation to keep promises and pay debts is not depen- 
dent on the possibility of our being found out if we disobey 
the rules. But we are asked to say what would be the case in 
a world in which moral rules had no sanction apart from each 
man's conscience. The only possible reply to this question is 
that in such a world we should have no use for the concepts 
of rules or obligation at all. For, although it is not the case 
that people who obey rules always do so from a fear of the 
consequences of disobedience, the concepts of a rule and an 
obUgation to obey it could never have arisen apart from that 
of a penaky for disobedience. The very word ' obligation ' be- 
trays this fact. In a moral Utopia men might always do the 
things that are in this world enjoined by our moral rules, but 
they would do them for some reason other than that of regard- 
ing them as obligatory. For, even if it is true that in our world 
a man's conscience is the only judge of his conduct, it is diffi- 
cult to see how people could have arrived at the duahstic 
notion of 'me and my conscience' unless they were familiar 
with judges of a very different sort, a sort which could not 
have existed in the world depicted by Glaucon and Adeimantus 
since they could not have carried out their functions. 

Part of the force of Plato's story Ues in the fact that Glaucon 
and Adeimantus are made to assume that the lucky owner of 
the ring would indulge in a Hfe of unbridled sensuality and 
selfishness. Once again, under the guise of an analysis of the 
logic of moral words, there is smuggled in a strange assump- 
tion about the sorts of things that people like doing and would 
do if the sense of duty did not restrain them. 



But suppose that it is admitted that I should in fact recog- 
nize the obligation even in the desert-island case. What fol- 
lows ? Simply that my sense of duty is exceedingly powerful, 
more powerful than I thought it was and so powerful as to 
overcome any pro-attitude that I might have towards breaking 
the rule. I have, after all, been trained from my earliest youth 
to keep promises; and it is not unnatural that I should have 
acquired a strong repugnance to breaking them. " What makes 
him think it right to act in a certain way is the fact that he has 
promised to do so, that and usually nothing more." There is 
an irony in the word ' makes ' that, I think, Ross did not intend. 

The fact that a man has made a promise can no more explain 
why he keeps it than any other fact. He may keep it because 
he has been so thoroughly trained that it does not occur to 
him to break it ; and in this case there is no need to ask for his 
motive, as he hasn't any. But if he chooses to keep the promise 
his choice requires to be explained in terms of a pro-attitude; 
and this may well be, not a pro-attitude towards the conse- 
quences, but a direct pro-attitude towards keeping a promise 
as such, a horror of or repugnance towards promise-breaking. 

It is not easy to describe the peculiar feeling called moral 
repugnance that in fact restrains us from breaking a moral 
rule even in cases where we think it right to do so. But, 
although it is impossible to describe it to someone who has 
never felt it, it is unnecessary to describe it to someone who 
has. If the reader is puzzled to know what I mean, let him try, 
for example, stealing something under conditions in which it 
is highly improbable that he will be caught. He will, I think, 
find it exceedingly difficult, if not impossible, to bring himself 
to do it. Such is the force of moral training. And he will also 
find it exceedingly difficult to tell a lie, even in a case where 
he is convinced that he ought to do so. We might paradoxi- 
cally, but not unfairly, say that in such a case it is difficult to 
resist the temptation to tell the truth. We are the slaves of our 
own consciences. 

But what is the connexion between the fact that I feel this 
special moral repugnance towards breaking a moral rule which 



I have been brought up to obey and the question whether I 
ought to break it? What, in Butler's terminology, is the con- 
nexion between the pozoer and the authority of conscience? 
Butler thought that its authority was 'manifest', a contention 
I shall examine in the next chapter. What is manifest, at least 
in some people, is its actual power, that is to say the fact that 
people find it very difficult not to become slaves of their con- 
sciences. " Having read my Flume I should become a Thrasy- 
machus."^ I doubt if it is as easy as that to become a Thrasy- 
machus even if it were true - which it is not - that a man who 
had no sense of duty would necessarily wish to become one. 
What the desert-island argument proves, if its conclusion be 
accepted, is not the fact that duties are intuited and indepen- 
dent of purposes, but that (a) the language of duty cannot be 
translated into the language of purpose, and (b) that moral 
habits die hard. 

I. Raphael: op. cit., p. 94. 





In defending the traditional, teleological approach to the 
understanding of moral language I may seem to have made 
so many concessions as to have abandoned the case. I have, 
for example, denied that deontological words can be defined 
in teleological terms and admitted that there are no logical 
limits to the possible objects of pro-attitudes other than the 
logical limits of language itself. It follows from this that a man 
might have a direct pro-attitude towards obeying a moral rule 
or towards doing what he thinks his duty, as such; he may 
have no desire whatever to do the action apart from its being 
his duty and no thought of the consequences, for himself or 
others, that the action is likely to produce. Now this motive 
is exactly v/hat we mean by the Sense of Duty and the man 
who acts on it is called conscientious. 

And is not to admit all this precisely to admit the truth of 
the deontologist's case ? He would agree that it is a truism to 
say that every man must have a motive for what he does; 
since, if there were no motive, it would not count as ' doing ' 
in any sense relevant to ethics. But he insists that the Sense 
of Duty is a very special motive; it is logically so different 
from other motives that it is a mistake to call it a pro-attitude 
at all; and it is morally peculiar in that, while a naan can ex- 
plain why he did what he did by reference to any motive, he 
can only justify his conduct when he acted from his sense of 
duty. Moral value, it is said, lies in conscientious action and 
in that alone. In this chapter I shall examine the claim of the 
sense of duty to be the only morally good, or at least the best, 

According to Kant an action has moral worth only if it is 



done from the sense of duty. As Professor Paton puts it : " An 
action done solely out of natural sympathy may be right and 
praise-worthy, but nevertheless it has no distinctively moral 
worth. "^ How is this to be interpreted? It is allowed that an 
action done from sympathy may have some 'worth', but not 
'moral worth'. It might, therefore, be suggested that Kant 
and his followers are deHberately using words in a special v/ay 
of their own ; but if this is so they meet with the usual fate of 
philosophers who recommend changes in ordinary usage. 
Questions which are either senseless or admit of an obvious 
answer when words are used in the ordinary way may become 
debatable when they are used in the new way. We might well 
agree that moral worth is superior to all other kinds of worth, 
for example that of talents and accomplishments and material 
goods. But this is because we normally think of moral worth 
as meaning the worth of any virtuous motive and we normally 
think of sympathy and benevolence as virtuous motives. But if 
moral worth is limited by definition to the worth of conscien- 
tiousness it is no longer so obvious that it is superior to all 
other kinds of worth. 

But I do not think that this interpretation is correct. Kant 
claims to be elucidating the value-judgements of the ordinary 
man. And, while it is a tautology to say that only virtues have 
moral worth (for moral worth just is the worth of virtues), 
it is not a tautology that conscientiousness is the only virtue 
and hence the only bearer of moral worth. Nor is it even true 
that ordinary men regard conscientiousness as the only virtue ; 
almost all men would include altruism in the list ; and ' altru- 
ism' means, not 'doing good to others for duty's sake', but 
'doing good to others for its own sake' or 'doing good to 
others for the sake of doing good to others '. 

Kant in fact slides from an opening sentence in which he 
uses the phrase 'good will' to mean any virtuous motive to 
his full-blown theory that the sense of duty is the only virtuous 
motive. His first point is to prove that a good will is the 

I. The Moral Law, p. 19. See also The Categorical Imperative, 
chapter III. 



only thing that is good without qualification; and he tries to 
prove this by showing that other good things, talents, quali- 
ties of temperament, gifts of fortune, and even the classic 'vir- 
tues ' of moderation and self-control, can sometimes be bad. 
"The very coolness of a scoundrel makes him, not merely 
more dangerous, but also immediately more abominable in 
our eyes than we should have taken him to be without it."^ 

All this may be true; but it is not to the point. For what 
Kant has to show is, not that non-moral assets and even some 
virtues are not good without qualification, but (a) that con- 
scientiousness is good without qualification and (b) that no 
other virtue is. 

The tacit equation between conscientiousness and moral 
virtue comes out well in Paton's treatment of the question 
whether moral virtue is the "highest" good.^ He contrasts 
conscientiousness with non-moral goods, such as artistic acti- 
nty and knowledge; but he does not even raise the question 
whether conscientiousness^is 'higher than' other moral vir- 
tues. Nevertheless it seems that this is an open question. 

And it is also an open question whether conscientiousness 
itself is good without qualification. Many of the worst crimes 
in history have been committed by men who had a strong 
sense of duty just because their sense of duty was so strong. 
[ should myself have no hesitation in saying that Robespierre 
would have been a better man (quite apart from the question 
af the harm he did) if he had given his conscience a thorough 
rest and indulged his taste for roses and sentimental verse. 
There is a story of an Oxford don who disliked Common 
Room life and whose presence caused himself and others acute 
distress. Yet he attended Common Room assiduously because 
be thought it his duty to do so. He would have done better 
to stay at home. 

In answer to this type of criticism Paton says: "It is cer- 
tainly true that good men may do a great deal of harm; and 
:his harm may spring, not from ofHciousness and vanity (which 

1. Trans. Paton: The Moral Law, p. 62. 

2. The Categorical Imperative, p. 42 



belong to moral badness) but from mere silliness and stupi- 
dity."^ But may not the harm also spring from their very 
conscientiousness? We might adopt the moral principle that 
conscientiousness is so valuable that a man ought to be con- 
scientious no matter what harm he does ; but it is quite another 
thing to say that their conscientiousness is never the source 
of the harm that good men do. 

Nor, I think, is the principle of the supreme value of con- 
scientiousness one that we have any reason for accepting. Its 
claim seems to rest partly on a confusion to be examined in 
section (3) and partly on the assumption that non-conscien- 
tious action must be both impulsive and selfish.^ 


The Value of Moral Virtues. "Virtues and vices are dispositions 
to behave in certain ways; but the names of virtues and vices 
do not merely designate those modes of behaviour. They are 
terms of praise and censure to such an extent that it would be 
logically odd to call a man ' brave ' or ' honest ' without intend- 
ing to praise him. We must now examine the question why 
some modes of conduct are so universally praised and others 
so universally condemned that their very names always carry 
a praising or condemning force unless it is expressly or by 
implication withdrawn. 

I shall try to show that the value of virtues is always an 
artificial value; but it does not follow that the modes of con- 

1. Op. cit., p. 40. 

2. It also rests on a confusion between acting on principle and acting 
for duty's sake. It is plausible to contrast acting on principle with acting 
on impulse and to suggest that every motive belongs to one or other of 
these types. But I think, with Paton, that Kant meant by acting for the 
sake of duty "acting out of reverence for the moral law"; and this, 
though compatible with, must not be identified with acting on prin- 
ciple. A man who acts on the principle of beneficence does not act out 
of reverence for any law, moral or otherwise ; though, if the moral law 
enjoins beneficence, he does what the law enjoins. 



duct that virtue-words designate are never natural. For we 
must distinguish between the question whether it is natural 
to behave in a certain way and the question whether it is 
natural to praise those who act in that way. The phrase ' natural 
virtue ' is, in fact, ambiguous. It can either mean ' mode of con- 
duct that is natural and also praised ' or it can mean ' mode of 
conduct which is naturally praised.' I shall use the phrase in 
the latter sense and give reasons for supposing that there are 
no such modes of conduct. 

A man may have a natural disposition to be beneficent, that 
is to say a natural pro-attitude towards the good of others, and 
he is called ' benevolent ' if this is so strong that he regularly 
acts for the sake of doing good to others. No doubt this pro- 
attitude is naturally stronger in some men than in others, and 
it can be strengthened or weakened by education. But it does 
not follow that benevolence is a natural virtue. For to say this 
is to say, not that it is natural to be benevolent, but that it is 
natural to praise the benevolent man. Similarly, to say that 
courage and honesty are natural virtues is to say, not that 
men are naturally bra^^e or honest, but that it is natural to 
praise the brave and honest man. 

Now it obviously cannot be proved that the value of moral 
virtues is always artificial. To prove this empirically we should 
have to examine the pro-attitudes of that well-known abstrac- 
tion, the Natural Man without Society. Nor could it be proved 
a priori, since there are no logical limits to the possible objects 
of pro-attitudes. But it is a maxim in philosophy that we should 
never assume anything to be part of the original constitution 
of human nature if any cause can be assigned to it ; and in this 
case a cause is not far to seek. There is an obvious connexion 
between the modes of conduct that men praise and the modes 
of conduct that they believe to bring about consequences 
towards which they already have a pro-attitude. 

Since praising and condemning are things that we do, it is 
necessary that we should have some motive for doing them; 
and we should not conclude that a motive is a natural one if 
it is possible to account for it in some other v/ay. Now it is 



tautological to say that every man has a pro-attitude towards 
other people doing good to him, since ' doing good to' means 
'bringing about those things towards which he has a pro- 
attitude'. And, though not tautological, it is obviously true 
that most men have a natural pro- attitude towards other 
people's doing good to third parties, not perhaps to all other 
men, since such widespread sympathy is very rare, but at least 
their family, friends, and neighbours. Most men are not misan- 
thropes, and in order to understand how artificial pro-attitudes 
arise it is essential to remember that we have to do with the 
majority, not with the exceptional case. 

Now benevolence is the desire to do good to others, and it 
is fairly obvious that men are more likely to succeed in doing 
good to others if they try to do so than if they try to do harm. 
We could easily imagine a world in which this was not so, a 
world in which people were so stupid and inefficient that the 
more good they tried to do the more harm they did. And, if 
we lived in such a world, benevolence would be a vice and 
malevolence a virtue. That this sounds odd is due solely to the 
facts of the world in which we live. It does not sound so odd 
in the case of some other virtues. 

Although some modes of conduct are universally praised, 
there are others which are regarded as virtues in some socie- 
ties but not in others. And the reason in every case for regard- 
ing something as a virtue is that it is believed to be beneficial 
in the special social and economic conditions of the society 
concerned. Thrift, for example, is valuable in some economic 
conditions and harmful in others; consequently we find it 
sometimes praised as a virtue and sometimes condemned as a 
vice. As the world becomes more closely knit, so that the 
prosperity of one country comes to depend more and more on 
the prosperity of others, patriotism ceases to be a virtue ; and 
we either cease to praise the patriotic man or change the 
descriptive content of the word 'patriotism'. 

In saying that moral virtues are not natural objects of pro- 
attitudes I do not intend to say that these pro-attitudes are 
never direct. Men do not always praise benevolence or courage 



because they have a pro-attitude towards its consequences in 
a particular case. A man may admire and honour courage in 
an enemy though this virtue is greatly to the disadvantage of 
himself and his friends; indeed we often praise a good man 
without any thought of the good consequences of his virtue. 
But the pro-attitude which such praise implies is analogous 
to an acquired taste. Although it may now be direct and spon- 
taneous, we have come to have it partly because we have been 
brought up to admire certain modes of conduct and partly 
because we recognize their instrumental value. 

The Janus-character of virtue-words is a fruitful source of 
misunderstanding here. For it tempts us to confuse (a) the 
meaning of a virtue-word, (b) the causes of our treating some- 
thing as a virtue, and (c) the motive for our so doing, (a) We 
use virtue-words as terms of praise, and a man who praises 
something need have no motive for praising other than a 
direct desire to praise it. (b) The cause of his praising a cer- 
tain mode of conduct may well be that it is customary to do 
so or that he has been trained to do so. We learn what modes 
of conduct to praise and what to condemn partly by being told 
and partly by imitating others; and it is easy to see how it is 
that modes of conduct often continue to be praised even when 
they are no longer thought to produce good consequences, 
(c) But the original motive for treating any mode of conduct 
as praiseworthy must either have been that men had a direct 
pro-attitude towards it or the belief that it has, in general, con- 
sequences towards which we have a pro-attitude. For it is 
logically odd to encourage people to try to do things to which 
we are opposed. 


Is conscientiousness necessarily the best motive? To say that con- 
scientiousness is a good motive or a virtue is, among other 
things, to praise the conscientious man and to encourage 
people to be conscientious ; and this is not to comment on the 



use of moral language but to make a moral judgement. It 
would therefore be very strange if it were logically necessary s 
that conscientiousness should be the best motive. It is, how- 
ever, worth while investigating this question, since so many 
philosophers have supposed that, while the value of other ; 
virtues is contingent, conscientiousness is necessarily 

Sir David Ross uses the following argument to prove that 
we must regard a man who acts from a sense of duty as a 
better man than one who acts from any other motive. " Sup- 
pose that some one is drawn towards doing act A by a sense 
of duty and towards doing another, incompatible, act B by 
love for a particular person. Ex hypothesi, he thinks he will 
not be doing his duty in doing B. Can we possibly say that 
he will be acting better if he does what he thinks not his duty 
than if he does what he thinks is his duty? Evidently not. 
What those who hold this view mean by 'acting from the 
sense of diity' is obeying a traditional, conventional code 
rather than following the warm impulses of the heart. But 
what is properly meant by the sense of duty is the thought 
that one ought to act in a certain way. . . . And it seems clear 
that when a genuine sense of duty is in conflict with any 
other motive we must recognize its precedence. If you 
seriously think that you ought to do A, you are bound to 
think you will be acting morally worse in doing anything 
else instead."^ 

It should be noticed that Ross has loaded the scales in favour 
of the sense of duty by representing the only alternative 
motives as impulses, a word which suggests that they are 
sporadic, wayward, and capricious.^ But sympathy, benevo- 
lence, patriotism, and ambition are not necessarily impulsive. 

1. The Right and the Good, p. 164. 

2. Ross in fact goes on to describe instinctive affection as 'wayward 
and capricious'. But, though affection may be instinctive it can be cul- 
tivated and need be neither wayward nor capricious. Ross's view seems 
to be another example of a mistake due to thinking of all motives other 
than the sense of duty as 'desires', 'inclinations', or 'impulses'. 


A man can consistently adopt a policy of doing good to others, 
not because he regards it as his duty, but because that is what 
he most wants to do or enjoys doing. The word 'wants' is, of 
course, far too weak a word to cover the pro-attitude of the 
non-conscientious altruist. But his altruism is not necessarily 
less consistent or more easily shaken than that of the man who 
tries to do good because he thinks it his duty. 

Secondly there is an ambiguity in the phrase 'acting better'. 
If this means that what he does is better in some non-moral 
sense, for example that it brings about better consequences, 
it is extremely doubtful whether a man who acts from a sense 
of duty in fact 'acts better' than one who does not. But if by 
'acting better' Ross means that a critic would necessarily re- 
gard the man who acts from a sense of duty as a morally 
better man, the argument begs the question. 

Indeed the passage I have quoted is mostly an appeal to the 
self-evidence of the proposition that a man who acts from a 
sense of duty is a better man than one who acts from any other 
motive. It is only in the last sentence that an argument is used 
to support this view; and the argument seems to depend on 
a confusion between what an agent necessarily thinks about 
his own action and what a critic or spectator necessarily thinks., 
Ross's object is to prove that Jones necessarily regards Smith 
as a better man if he does what he (Smith) thinks he ought 
to do; but the statement at the end of the quotation is only 
true if 'you' is taken to refer to the same person throughout. 
We must distinguish the following three statements: 

(i) I think that I ought to do A but that I would be a 
better man if I did B. 

(2) I think that you ought to do A but that you would be a 
better man if you did B. 

(3) You think that you ought to do A, but you would be a 
better man if you did B. 

Now there is an air of contradiction about (i) and (2), but 
not about (3). And the reason why (i) is logically odd is that 
'I ought to do A' expresses a decision to act in a certain way 



and implies that the decision is of a certain type, namely one 
based on reasons which, in a moral case, may take the form 
of a belief that A would be fitting or in accordance with a cer- 
tain moral rule. A man who said that he ought to do A but 
would be morally better if he did B is in the same breath de- 
ciding to act on a moral principle and condemning himself for 
making this decision. But to condemn himself is to abandon 
the moral principle in question. 

And (2) is logically odd for a similar reason. To say "you 
ought (morally) to do A" is to advise a man to adopt a certain 
moral principle and the force of "But you would be a better 
man if you did B" is to retract this advice. It is as inconsistent 
to recommend and to condemn a moral principle in the same 
breath as it is to decide to adopt and to condemn a moral 
principle in the same breath. 

But (3) is not logically odd at all; it is the natural way for 
Jones to express his moral disagreement with Smith. Now 
conscientiousness is an extremely valuable motive and it is so 
valuable that we often wish to encourage a man to be con- 
scientious even in a case in which we think that the principle 
on which he thinks he ought to act is a bad one. In such a case 
we might well wish to encourage him to do what he thinks 
right without wishing to endorse the principle on which he 
proposes to act. We should then say "I think you ought to do 
B ; but, if you are really convinced that you ought to do A, 
then you ought to do it. For what really matters is not that 
you should act on the right principle but that you should act 
on the principle that you believe to be right." But I do not 
think it is logically necessary that we should rate conscien- 
tiousness as highly as this nor that, as a matter of fact, we 
always do. Statement (3) is not logically odd except in the 
mouth of a man who has already accepted the very principle 
of the supreme value of conscientiousness which Ross is trying 
to establish. 



The Value of Conscientiousness. The case of conscientiousness 
differs from that of other virtues in two ways, (a) The mode 
of conduct that the word designates is an artificial mode of 
conduct ; and (b) the value of conscientiousness is artificial in 
a way in which that of other virtues is not, 

(a) to be conscientious is to do what one believes to be 
right, not for the sake of bringing about a certain result nor 
for the sake of doing what is done, as such, but for duty's sake. 
And two different motives can lead a man to do the same 
thing. For example a man may be both altruistic and con- 
scientious, and such a man will help a blind man across the 
road both because he wants to help him and also because he 
thinks it his duty to do so. But, in order to simplify the issue, 
I shall consider the case of a man who acts from the sense of 
duty alone and does something for no other reason than that 
he thinks he ought to do it and has a pro-attitude towards 
doing what he thinks he ought to do, as such. This certainly 
occurs; how has it come about? 

I have incurred a debt and I pay it with no desire to part 
with my money, no thought of the welfare of the recipient, and 
no expectation of gain. My sole motive is the desire to con- 
form to the moral rule 'Pay your debts'. It will hardly, I think, 
be argued that anyone has a direct, natural pro-attitude to- 
wards obeying this rule, or even the germ of such an attitude 
that could be fostered by education. It may be that men are 
naturally ritualistic, that they have an innate love of order- 
liness and doing things according to rule. Anthropologists are 
divided on this point; but in any case it is irrelevant, since 
conscientiousness is not the desire to conform to any rule, but 
the desire to conform to a rule which one regards as right. And 
it is therefore necessary to explain how I came to regard this 
particular rule, 'Pay your debts', as a right rule to adopt. 

As usual custom and education can explain much but not 
everything. In the case of any given man it is no doubt true 



that he adopts those rules which are customary in his society '' 
and to which he has been trained to conform. But these causes 
are not reasons and they cannot explain how the customary ; 
rules which a man has been trained to adopt came to be j 
accepted as the right rules. There must have been some motive | 
for establishing the rule in the first place and, in so far as men I 
are rational, there must be some reason for adhering to it. f 
Now the motive for adopting a rule cannot have been the sense 
of duty, since the sense of duty is the desire to do whatever 
is laid down by the moral rules we have adopted. A man who 
acts from a sense of duty pays his debts because he thinks it 
right to do so ; he must therefore have some reason for think- 
ing it right other than the fact that his sense of duty bids him 
do it. 

This argument is equally valid whether we think of the 
sense of duty as a desire to conform to ' objective ' moral rules, 
to a customary code, or to those rules which a man adopts for 
himself. In each case, if he adopts the rule he must have some 
motive for adopting it, and this motive cannot be a desire to 
conform to it. It must be a direct or indirect pro-attitude to- 
wards doing what the rule lays down irrespective of the fact 
that the rule does lay it down. 

(b) To say that conscientiousness is a natural virtue is to 
say that it is natural to praise (and therefore to have a pro- 
attitude towards) obedience to a moral rule. But this, although 
not logically impossible, seems very unlikely to be true. Here 
is a man who has incurred a debt and pays it for no other 
reason than that he thinks he ought to do so. Why should I 
praise and admire him rather than condemn and despise him ? 
These questions sound odd only because 'conscientious' is 
already a term of praise. If we are careful to exclude the prais- 
ing force and to think of it as meaning only ' acting from a sense 
of duty' it clearly makes sense to ask why we should praise the 
conscientious man. 

And the reason cannot be found either in a regard for our 
own interest or in a regard for that of others. I may or may 
not have a pro-attitude towards the redistribution of wealth 



involved. This is immaterial, since the question is not 'Why 
do I approve of his paying the debt?', but 'Why do I approve 
of his doing this from a certain motive?'; moreover we ap- 
prove of conscientious action even in cases in which we have 
no pro-attitude towards any element in the situation other than 
the motive of the agent. 

Now the value of all good motives is, as we saw, artificial. 
We may come to have a direct pro-attitude towards the types 
of conduct designated 'virtuous' and praise the virtuous man 
without any thought of the consequences of his action in this 
particular case ; but the type of conduct concerned would never 
have come to be called virtuous if it was not believed to have 
good consequences. But the value of conscientiousness is arti- 
ficial in another way also. For conscientiousness is not the dis- 
position to do certain sorts of things that are, in fact, valuable, 
but the disposition to obey certain rules; and its value there- 
fore depends on the value of the rules, which are themselves 
artificial devices for ensuring certain states of affairs that we 
wish to ensure. 

Now to be conscientious is not to conform to an accepted 
moral code, but to conform to rules to which the agent himself 
thinks he ought to conform. But, although it is possible for 
some individuals to adopt rules that conflict with the accepted 
code, it is logically necessary that such cases should be rare. 
There could be no such thing as an accepted code if most 
people did not accept it. It follows therefore that, although 
there may be exceptions, in the majority of cases a conscien- 
tious will do those things that are laid down in the 
accepted code more often than a non-conscientious man will; 
and since the code consists of rules which are believed to pro- 
mote the interests of society, it follows that a conscientious 
man must be more likely to do what is believed to be in the 
interests of society than a non-conscientious man. This belief 
may be false; but, even if it is false, it explains vv^hy people 
are praised for being, and encouraged to be conscientious even 
in cases in which we do not endorse the rule which they adopt 
and deplore the consequences of their actions. 



The Unique Position of Conscientiousness. Apart from the bad 
reason provided by the dogma that all non-conscientious action 
is impulsive or selfish, there is a good reason for allowing con- 
scientiousness a special place on the scale of moral virtues. A 
man who displays some other virtue, for example courage or 
honesty or generosity, can be relied on to do just those things 
that belong to his special virtuous disposition; and these vir- 
tues can only be exercised in comparatively narrow ranges of 
situations. But conscientiousness is a substitute for all other 
virtues, and its unique value lies in this fact. The so-called 
'natural virtues' are dispositions to do certain sorts of things 
towards which we have, in general, a pro-attitude; and moral 
rules are rules enjoining these same things. Hence the con- 
scientious man will do exactly the same thing that a man who 
has all the natural virtues will do. He does not do them for 
the same reason ; and he is not brave or honest or kindly, since 
he acts for the sake of doing his duty, not for the sake of doing 
the brave, honest, or kindly thing. But he will do what the 
brave, honest, and kindly man does. 

The value of conscientiousness is therefore not unlike that 
of money. Just as a pound note has no instrinsic value but is 
valuable because it can be used to buy any of a large range of 
goods, so the desire to do his duty, whatever it may be, will 
lead to a man's doing any of a large range of valuable actions. 
And the value of conscientiousness is like that of money in an- 
other way also. Just as a pound note is valueless except in a 
country where it is accepted in return for goods, so many of 
the duties of 'special obligation', for example promise-keeping 
and debt-paying, are only valuable in a society in which the 
rules enjoining them are generally obeyed. The ends which 
these rules are designed to promote would not be promoted 
by obedience to them unless there was a general system of 
obedience, so that people could be relied on to keep their 
promises and pay their debts. Without such a system the very 
notions of a ' promise ' or a ' debt ' would be unintelligible. 

To ask whether conscientiousness is the highest virtue is 
not unlike asking the question whether money is more valu- 



able than other goods. The answer depends on how much you 
have. Moreover this is a question the answer to which is a 
moral judgement and it cannot therefore be answered either 
by observation or analysis of moral language. Aristotle held 
that a man was not really good unless he enjoyed doing what 
is good, and I am inclined to agree. The sense of duty is a 
useful device for helping men to do what a really good man 
would do without a sense of duty; and since none of us belongs 
to the class of 'really good men' in this sense, it is a motive 
that should be fostered in all of us. But it plays little part in the 
lives of the best men and could play none at all in the lives of 
saints. They act on good moral principles, but not from the 
sense of duty; for they do what they do for its own sake and 
not for the sake of duty. 





At the end of chapter i6 I suggested that the actual power 
of conscience was more obvious than its authority and that a 
man might be a slave to his own conscience; and in chapter 12 
I said that it was logically odd to say 'This is the morally 
better course; but I shall do that'. These paradoxes must now 
be explained. 

It might seem to be tautologous to say that a man ought 
to do what his conscience tells him to do ; for is not the Voice 
of Conscience precisely the voice that tells him what he ought 
to do ? But this argument is plainly specious ; for a decision to 
do something aever follows logically from a command to do 
it. 'I ought' never follows from 'You ought', and if Conscience 
is described - as it is both in philosophical literature and com- 
mon speech - as a voice that tells you what you ought to do, 
its function is that of advising, exhorting, or commanding, not 
of deciding or choosing. 'Everyone ought to obey his con- 
science ' is a general moral commandment issued to everyone, 
including the speaker ; it is not a logical truism. 

Nor can this conclusion be evaded by saying that the voice 
of conscience is infallible, that 'you ought' entails 'I ought' in 
the special case in which it is conscience that issues the com- 
mand. For if this were so we should have to say that men can 
mistake some other voice, perhaps that of the Freudian Father- 
Substitute, for that of conscience ; and we should have no way 
of distinguishing the true conscience from the false except by 
saying that the ' you ought ' is a genuine command of conscience 
only in cases in which it does entail 'I ought'. Nothing is more 
certain than the fact that the consciences of different men con- 
flict; and, even if it were true that Jones necessarily thinks 



Smith a better man if he follows his conscience (which I have 
given reasons for doubting), it is certainly untrue that Jones 
always thinks that Smith's conscience has given the right 

The philosophical tradition that treats conscience as an in- 
ternal judge is partly responsible for the theory that conscience 
has "manifest authority"; for judges are notoriously people 
who have authority. If conscience is the court from which there 
is no appeal, it is tautological to say that it is right for a man 
to obey his conscience, since conscience is, by definition, the 
authority competent to judge what is right and what is wrong. 
But v/e must not confuse the office of the judge with that of 
the advocate. The role of the former is to pronounce a verdict, 
that of the latter to plead a cause. 

The confusion is reached in the following way. *I ought' is 
used to express a verdict or decision. It differs from 'I shall' 
in that, while 'I shall' can be used to express any decision, 'I 
ought' is only used to express decisions of a certain kind, 
namely those based on rules. It is for this reason that it is logi- 
cally odd to say "I know I ought to do X, but shall I do it?". 
Unless ' shall I ? ' is being used in a predictive sense to be con- 
sidered later, a man who says it has not yet reached a decision 
and cannot, in consequence, say 'I ought' in the judicial, ver- 
dict-giving sense. 

But 'I ought' is also used, not to express a decision, but in 
the course of making up one's mind before a decision has been 
reached. A man may hesitate between two moral principles 
and say to himself at one time ' I ought to do X' and at another 
'But on the other hand I ought to do Y' or he may contrast 
*I ought' with 'I should like to'. In the first of these cases he 
is hesitating between two moral principles, in the second be- 
tween acting on a moral principle and acting on some other 
motive. But in neither case has he arrived at a verdict. In the 
first case it is quite natural to represent the two 'oughts' as 
being spoken by internal moral authorities advising or telling 
him what to do ; and in the second to represent the conflict as 
one between the Voice of Conscience and Desire. But these 



are the voices of advocates, not of judges; and what they say 
is, not 'I ought', but 'you ought'. 

The difference between 'you ought' and 'I ought' is ob- 
scured by two facts, one empirical and one logical. In the first 
place moral struggles are comparatively rare; we have often 
only to recognize the ' you ought ' of conscience to pass imme- 
diately to the ' I ought ' of decision. (Whether we carry out the 
decision is another matter; we may lack self-control.) And" 
secondly our talk about moral judgements, as opposed to the 
expressions we use in making moral judgements or decisions, 
is always put in indirect speech. And here both the verdict- 
giving ' I ought ' and the self-hortatory ' you ought ' become ' he 
ought'. In a description of a man making up his mind about a 
course of action "He thinks that he ought to do X" might 
mean "He is saying to himself 'You ought to do X'", i.e. he 
is telling himself what to do. But it might also mean "He is 
saying to himself 'I ought to do X'", i.e. he has arrived at 
a decision about what to do. 

Duty and Inclination. This way of representing deliberation 
as a conflict between voices or forces is not wholly unnatural. 
But if either the metaphor of the council chamber or that of the 
conflict of mechanical forces is taken too seriously it leads to 
highly paradoxical results. According to some psychologists 
the Voice of Conscience in an adult is the ghost or memory of 
the father who told him what to do when he was a child and 
punished him for not doing it. Now so long as psychologists 
confine themselves to describing and explaining empirical 
phenomena and do not draw moral conclusions, there is no 
reason why a philosopher should quarrel with them ; and if he 
does, he ought to produce evidence that rebuts the psycholo- 
gists' explanation. But philosophers often object to this account 
of the genesis of conscience on the grounds that it ignores the 
peculiar moral authority of conscience. They claim rightly that 
we are under no necessary obligation to do what the voice of 
conscience tells us if this voice is really what the psychologists 
say it is. But in their account of decision and obligation they 



tend to reproduce the very feature that makes the psycholo- 
gist's account irrelevant to the question whether I ought to 
obey my conscience. Conscience, they say, is a special non- 
natural voice that speaks with authority; but they still repre- 
sent it as a voice which issues orders. And if this is what it is 
it will always make sense for me to ask whether I ought to 
do what my conscience tells me to do. For a moral decision is 
a decision to act on a principle that one freely accepts, not a 
decision to act on a principle on which one is told to act. The 
' voice ' or ' force ' to which conscience is likened in the council 
chamber and mechanical metaphors is one of the participants 
in the contest and cannot be identified with the person who 
decides between the participants. To make a moral decision 
is neither to be ordered (though it may be a decision to obey 
an order) nor to be the prize of a victorious inner force 
(though it may be a decision to follow a certain inclination); 
it is to decide; and this is something that / do, not something 
that is done by voices or forces inside me. 

Some philosophers will object to this account of the differ- 
ence between the self-hortatory 'you ought' and the verdict- 
giving 'I ought' on the grounds that, in the special case of 
conscience, the two are identical. My conscience, they will say, 
is myself. While they are prepared to talk of desires or inclina- 
tions as internal forces which operate on 'me', the 'self or 
*self-acting-in-accordance-with-conscience' is, as it were, the 
billiard-ball on which these forces act. The only difference 
between the case of the billiard ball and that of the self is that, 
whereas the behaviour of the billiard ball is completely deter- 
mined by the forces acting on it, the self is capable of spon- 
taneous action. I choose freely only when I am not obliged by 
my desires or inclinations but do what my 'self decides to do. 

But this theory seriously distorts our account both of choos- 
ing and of responsibility. I shall consider its application to 
responsibility in the next chapter and confine myself here to 
the suggested analysis of choice. Moral conflict is now repre- 
sented as a battle between 'me' (or my 'self or 'my con- 
science') and 'my desires'; and if this is so it is nonsense to 



ask whether or not I can choose to overcome a particular 
desire. For 'I' am now represented as one of the participants 
in the conflict. Either I win or I lose. If I win, I act freely 
because my action is that of the self-propelled conscience or 
conscience-propelled self; if I lose, I do not act freely, because 
the action is that of a desire which is not me at all. But in 
neither case can I choose between what I ought to do and 
what I want to do. 

I have already suggested that it is paradoxical to represent 
all motives other than the sense of duty as ' forces ' which oblige 
me to act as I do, since this entails that I do not act freely when 
I do what I want to do ; and I suggested that a worse paradox 
was to follow. It is this. There can be no such thing as inten- 
tional or even voluntary wrong-doing, and therefore no such 
thing as just blame or punishment. In the mouth of a Socrates 
or a Spinoza there would be nothing strange about this con- 
clusion; for Socrates thought no wrong-doing could be 
voluntary and was puzzled to know how any man could de- 
serve blame, and Spinoza was prepared to push the theory to 
its inevitable conclusion and say that blame is never justified. 
A wise man tries to understand why men behave as they do; 
only a fool blames them. 

But the theory of the self-propelling conscience is often 
found in conjunction with the view that conscientiousness is 
the only virtue and acting against one's conscience the only 
vice. And it is this combination that is paradoxical, since on 
this theory a conscientious action is the only type of free 
action, all actions prompted by desire being unfree. If con- 
science wins the day I act freely and am good; if desire wins 
the day, I am bad but / do not choose to do what I do. Now 
all this may be true ; but, if so, ordinary men have for centuries 
been labouring under a profound delusion. For nothing is more 
certain than that they believe that a man can choose to do what 
is wrong and that he cjiooses in exactly the same sense of 
'choose' as he does when he chooses to do what is right or 
when he chooses to do something in a case in which morality 
is not involved. No doubt there are many diflPerences between 



choosing the path of duty and choosing a place for a hoKday; 
but it is paradoxical to suggest that we have to do with dif- 
ferent senses of the word 'choose'. 


Wickedness and Moral Weakness. We normally distinguish 
between two differeiit types of wrong-doing; (a) cases in 
which a man adopts and adheres to bad moral principles, and 
(b) cases in which he has good moral principles but fails, on 
some occasion, to live up to them. The latter is moral weak- 
ness; for the former we have no unambiguous name and I 
shall call it 'wickedness', although this word is too strong 
for niinor defects of this kind. The distinction is one of 
principle, not of the importance or gravity of the offence. 
What is common to both cases is that a man does something 
wrong, something, that is, that a spectator would condemn; 
the difference lies partly in the fact that the morally weak 
man condemns himself, while the wicked man does not. 

We sometimes refer to what I have called wickedness as 
'vice'; but we also use the word 'vicious' in respect of defects 
that, however much we may regret them and want to alter 
them, we do not regard as morally reprehensible. Addiction 
to opium is a vice, as is also any bad habit that a man cannot 
break however hard he tries. But these are not culpable 
states simply because, whatever may have been the case in 
the past, he cannot now avoid them. We shall see later that 
the concept of 'trying' is a difficuh and important one and 
that it is difficult in practice to discover whether or not a man 
eould have overcome his vicious habit or craving. But if he 
cannot, his vicious condition is one to be remedied by educa- 
tion, medical treatment, psycho-analysis or whatever means 
the wit of man can devise; it does not call for moral con- 

Now the theory considered at the end of the last section 
seems to me to make two profound mistakes. Since it treats 



all wrong-doing as succumbing to temptation, that is to say 
doing what a man knows that he ought not to do, it altogether 
ignores wickedness. And it distorts the nature of moral 
weakness by taking the case of the drug-addict as a model. 
The theory represents all motives other than the sense of 
duty as if they were cravings, that is to say 'external' forces 
which compel a man to act against his own moral principles; 
and drug-addiction is the model because this is indeed a case 
in which we do think of a desire or craving as an external 
force. ' External ' is, of course, a metaphor ; but the metaphor 
has point, since we use about the addict the same sort of lan- 
guage that we would use about a man who is physically com- 
pelled. 'He couldn't help it'; 'the craving was too strong for 
him'; 'he had no choice'. 

But if the weak-willed man really deserves blame, drug- 
addiction is not a good model to use for understanding his 
condition ; and if we represent desires as forces we shall have 
to find a criterion for deciding whether or not the craving was 
so strong that he could not have resisted it. In the next chapter 
I shall examine a theory that tries to do this. But the most 
paradoxical feature of the theory lies, not in the use of this 
inappropriate model, but in the complete neglect of cases in 
which it does not even begin to look appropriate. These are 
the cases of wickedness, cases in which a man, so far from 
struggling with temptation, neither tries nor thinks that he 
ought to try to do anything other than what he does. He 
may, of course, know that he ought not to do what he does 
in the sense that he knows that the practice is morally con- 
demned by others. But he does not believe that he ought not 
to do it in the verdict-giving sense of * ought ' ; on the contrary 
his action is an expression of the moral principles that he es- 
pouses. Now this is precisely the condition of the deliberately 
dishonest, cowardly, mean, or callous man, and to neglect 
these cases or to pretend that such men really know better but 
are continually giving in to temptation is to make nonsense 
of most of our moral judgements. 

Can we really say that the poisoner who waits years for his 



opportunity and plans every move with the greatest care does 
not act freely or that he has good moral principles but fails, 
through temptation, to live up to them ? If a man consistently, 
over a long course of years, tries to get the better of his 
fellows in all the transactions of daily life or if he is never 
moved by the consequences of his actions for other people, 
we might say colloquially that 'he has no moral principles'. But 
this clearly means, not that he literally has no moral principles 
or that he has good ones and continually succumbs to tempta- 
tion to act against them, but that he has bad moral principles. 
And it is surely absurd to say that, while it might be expe- 
dient to restrain him, it would be unjust to blame him. Yet 
this is precisely what we do say about drug-addicts. The 
drug-addict has to be dealt with, but he escapes moral censure 
just because he does not choose to do what he does. 


'/ ought' and '/ shalV. I have throughout treated 'What shall 
I do .? ' as the fundamental question of ethics and tried to show 
that moral concepts can only be understood by relating them, 
often in very indirect and complicated ways, to the concept of 
decision. But it might be objected that the fundamental ques- 
tion is not 'What shall I do.?' but 'What ought I to do?' 
and the fundamental concept not decision but obligation. My 
reason for treating the 'shall' question as fundamental is that 
moral discourse is practical. The language of 'ought' is intel- 
ligible only in the context of practical questions, and we have 
not answered a practical question until we have reached a 
decision. ' I shall do this ' is the general formula for expressing 

Now ' I ought ' can, as we have seen, also be used to express 
a decision, a solution to a practical problem ; but when it is so 
used it is a special case of ' I shall '. For it is used, not to address 
a command or piece of advice to oneself, but to express a deci- 
sion, and it differs from ' I shall ' only in that it contextually 



implies that the decision is of a certain type, namely that it is 
based on reasons and, if it is a moral 'ought', that the decision 
is in accordance with my moral principles. If I say to myself ' I 
ought to pay that debt' in the verdict-giving sense of 'ought', 
I am not merely recognizing the existence of a moral rule; I 
am subscribing to that rule. Only so can we understand why 
there is no logical gap between 'ought' and 'shall'. 

In such a case, therefore, ' I ought' entails ' I shall', of which 
it is a special case ; and ' I ought, but I shall not ' is a contra- 
diction. But this contradictory character of ' I ought, but I shall 
not ' is obscured by the fact that ' shall ' also has another use in 
which it is not self-contradictory to say 'I ought, but I shall 
not' or logically odd to say 'I know I ought, but I wonder 
whether I shall'. This is the predictive use. 

"I know that I ought not to get angry with Jones; he really 
means well; but I expect I shall, because I always do find myself 
getting angry with him." "I ought to take a firm line with 
Brown ; but / shall probably give in to him in the end ; his smile 
is so irresistible^ "/ wonder whether I shall have the courage 
to ask for the rise that I know I deserve." 

In these cases the use of 'I shall' is not practical but pre- 
dictive. As the phrases in italics show, we are adopting the 
standpoint of an observer, predicting our conduct in the future 
from a knowledge of our conduct in the past. We suspect our- 
selves of weakness of will or lack of self-control ; for we know 
that, when the time comes, we do not always act on our own 

The very phrase 'self-control' shows that there is point in 
the traditional way of thinking of a man as consisting of two 
or more people, sometimes a better and a worse 'self ; and the 
phrases 'weakness of will', 'lack of self-control', and 'loss of 
self-control ' are used to describe cases in which a man fails to 
do or predicts that he will fail to do what he has made up his 
mind to do, cases in which action runs counter to choice. 

It is these cases that give some plausibility to the self-pro- 
pelling conscience theory; for in these cases a man identifies 
'himself with his moral principles; but it is a mistake to treat 



such cases as typical of all moral wrong-doing; that is exactly 
what they are not. Ordinary moral language recognizes the 
distinction between having bad moral principles and being 
morally weak, and ordinary moral practice reserves its severest 
condemnation for the former. But, as often, ordinary practice 
has no precise way of distinguishing these two cases or of dis- 
tinguishing the wicked man, the weak-willed man, and the 
addict; and ordinary language has no precise way of formulat- 
ing the criterion or explaining why we make the distinctions. 
It is the search for such a criterion that gives rise to the philo- 
sophical problem of freewill. Perhaps, in the last resort, the 
weak-willed man cannot be distinguished from the addict and 
perhaps even the man of bad moral principles cannot help 
being what he is? It would be strange if these things were so; 
but there are truths that we all accept that appear to entail 
them, and this is one of the main ways in which philosophical 
problems arise. 



Freedom and Responsibility (i) 


We have now to consider the logic of the language which we 
use to ascribe responsibility, to award praise and blame, and 
to justify our moral verdicts. I shall consider five types oi 
moral judgement. 

He broke a law or moral rule. (i) 

He could have acted otherwise. (2) 

He deserves censure (or punishment). (3) 

It would be just to censure (or punish) him. (4) 

He is a bad (cruel, mean, dishonest, etc.) man. (5) 

It is clear that all these are logically connected. It is not 
just a fact about the world that we learn from experience that 
only bad men deserve blame or that it is only just to blame 
those who could have acted otherwise. Yet the items cannot 
all be treated as analytically connected ; for we should then find 
that it was senseless to ask certain questions that obviously 
do make sense. 

For example, the character- words used in (5) are partly 
descriptive; and it makes sense to ask whether a person who 
is consistently mean or dishonest deserves blame. To give, as 
most of us would, an affirmative answer would be to use, not 
to analyse, moral language. It would also be a mistake to say 
that it must (logically) be unjust to blame someone who could 
not have acted otherwise, on the ground that this is part of 
what 'unjust' means. For (2) is a theoretical statement, while 
'unjust' is a G-word contextually implying that no one 
ought to blame him. Nor does it help to say that we have in- 
sight into necessary synthetic connexions between the items 



on the list; for this is simply to say that we know them to be 
connected but cannot understand how. The connexions are of 
the quasi-logical kind that can only be understood by examin- 
ing the conditions under which the various expressions are 
used and the purposes of using them. 

I shall start by considering the connexions between (i), 
(3), and (4). The connexion between (3) and (4) seems to 
be analytic. If a man deserves blame, someone would be justi- 
fied in blaming him. Not necessarily you; for you may be in 
no position to cast the first stone or to cast any stone 
it all. 

Now ' punishment ' is a legal term and, in the case of punish- 
ment at least, (3) and (4) logically imply (i). A man can be 
justly punished only if he has broken a law, and the same 
ipplies, although naturally in a looser way, to moral censure. 
Fo deserve censure a man must have done something wrong, 
;hat is to say broken a moral rule. Now why should this be so ? 
rhis question has already been partly answered in chapter 16. 
Punishment ' is a complex idea consisting of the ideas of in- 
licting pain, on someone who has broken a law, in accordance 
A^ith a rule laying down the correct punishment. But we have 
Jtill to ask why we make use of this complex idea at all. Re- 
nembering that 'just' is a G-word, it is necessary to suppose 
:hat anyone who says that Jones deserves punishment must 
lave a pro-attitude towards his being punished. But why 
ihould we wish to encourage the infliction of pain on those 
vho have broken a law ? The classical utilitarian answer is that 
t will either reform the criminal or deter potential criminals 
)r both. Now, since laws and moral rules are devices for bring- 
ng about ends, we must have a pro-attitude towards reform- 
ng those who break them and deterring others. So, if it is a 
"act that punishment has these effects, this will explain the 
connexion between the infliction of pain and the breach of a 

But this simple theory will not do, if only because potential 
criminals would be as efficiently deterred by the punishment 
)f an innocent scapegoat who was beUeved to be guilty as by 



that of a guilty man ; and, whatever the effects might be, this 
would not be just. And we have also seen in chapter i6 how 
this simple theory can be amended. For we there saw that, 
although v/e might have a system of dealing with each situa- 
tion as it arose, there were great advantages in having legal 
and moral codes. And it is because we have these codes that 
neither the punishment of Jones nor an adverse moral verdict 
on him could (logically) be called 'just' unless he has broken 
a law. Without the code we could still recommend people to 
inflict pain on Jones to stop him doing what he does, but the 
peculiar force of 'just' could not be carried by any word. And, 
granted that we have rules, it is clear that the purpose of 
punishment and blame is relevant, not to the question ' Should 
Jones be punished or blamed ? ', but to the questions ' Should 
the sort of thing that Jones did be prohibited by a rule to 
which a penalty is attached ? ', 

The question ' What justifies punishment ? ' in fact conceals 
an ambiguity which is largely responsible for the dispute be- 
tween those who answer it in terms of retribution for crime 
and those who answer it in terms of deterrence and reform. 
If we have in mind the judge's problem, the utilitarian answer 
is clearly inadequate ; but if we are thinking about the legis- 
lator's problem, it seems very plausible. Each party has tried 
to extend their answer to cover both cases. But even if we are 
thinking about the legislator's problem, it would be an over- 
simplification to say that legislators either do or should decide 
what laws to have solely by reference to the purpose of having 
laws. There are two reasons why they do not do so, one bad 
and one good. The bad reason is that they are still to some 
extent in the thrall of the philosophical theory of Natural Law, 
v/hich itself confuses the judge's problem with the legislator's. 
But there is also a good reason. It is desirable (on grounds of 
utility) that the law should be consistent and stable and that 
the penalty laid down for one offence should not be wildly out 
of line with those laid down for others. Consequently, unless 
we aire to revise the whole legal code every time we make a 
new law, it is expedient not to consider the proposed law in 



isolation but to consider it as part of a system that we do not, 
on this occasion, wish to disturb. 

Just as we might, but do not, live in a world in which every 
case was decided by an omnicompetent judge without refer- 
ence to any general principle other than that of utility, so we 
might live in a world in which legislators decided what laws 
to pass solely by reference to this standard. But there are as 
good reasons for rejecting this system as there are for reject- 
ing the system of judges not boimd by laws. We know too 
little about the probable effects of any particular penalty and 
about the repercussions which a new law-cum-penalty is likely 
to have on other parts of the system. Hence even legislators 
do well to criticize proposed laws not only by reference to the 
purpose of having laws but by reference to the current system 
of laws, that is to say ' Justice '. The connexions between the 
justice of a punishment and its utility are thus exceedingly 
complex; but the fact that utilitarians have oversimplified 
them is a reason, not for abandoning their theory or retreating 
into the asylum of intuition, but for revising the theory. It 
cannot be an accident that the punishments we call 'just' on 
the whole tend to reform and deter. And if in a particular case 
we find that they do not serve these ends, we tend to amend 
the law. On the Natural Law theory it would only be right to 
amend a law if we discovered that it conflicted with natural 
law. How we discover this is in any case a mystery, and it 
v/ould be most remarkable if the discovery always went hand 
in hand with the discovery that the law fails to fulfil its pur- 


The most difficult and important of the items on our list of 
moral judgements is 'He could have acted otherwise' (2). 
The facts about its logical connexions with the others are 
tolerably clear. It is a necessary condition of all except (i) 
and it is also a necessary condition of (i) if 'He broke a law' 
is taken to imply that he broke it voluntarily. What is not 

K 273 


SO clear is what (2) means or why it should be a necessary 
condition of the other items. 

A man is not considered blameworthy if he could not have 
acted otherwise; and, although it is often easy to decide in 
practice whether he could have acted otherwise or not, it is 
not clear how we do this or why we should think it necessary 
to do it. Let us first examine the use of ' could have ' in some 
non-moral cases. 

' Could have ' is a modal phrase, and modal phrases are not 
normally used to make straightforward, categorical state- 
ments. 'It might have rained last Thursday' tells you some- 
thing about the weather, but not in the way that 'It rained 
last Thursday' does. It is sometimes said that it is used to 
express the speaker's ignorance of the weather; but what it 
expresses is not just this but his ignorance of any facts that 
would strongly tend to rule out the truth of 'It rained'. It 
would be a natural thing to say in the middle of an English, 
but not of a Californian summer. But, whatever it does express, 
what it does not express is a belief in a third alternative along- 
side ' it rained ' and ' it did not rain '. Either it rained or it did 
not; and 'it might have rained' does not represent a third 
alternative which excludes the other two in the way that these- 
exclude each other. 

But these modal phrases are also sometimes used in cases 
in which they cannot express ignorance since they imply a 
belief that the event concerned did not occur. It would be 
disingenuous for a rich man to say ' I might have been a rich 
man ' ; but he could well say ' I might have been a poor man ' 
while knowing himself to be rich. The puzzle here arises from 
the fact that, if he is rich, he cannot be poor. His actual riches 
preclude his possible poverty in a way that would seem to 
imply that we could have no use for ' he might have been poor '. 
But this is only puzzling so long as we try to treat these modal 
expressions in a categorical way. 

' Would have ' and ' might have ' are clearly suppressed hypo- 
theticals, incomplete without an 'if . . .' or an 'if . . . not . . .'. 
Nobody would say ' Jones would have won the championship ' 



mless (a) he believed that Jones did not win and (b) he was 
)repared to' add 'if he had entered ' or ' if he had not sprained 
lis ankle ' or some such clause. 

It is not so obvious that ' could have ' sentences also express 
lypotheticals ; indeed in some cases they obviously do not. If 
. man says ' It could have been a Morris, but actually it was 
n Austin', it would be absurd to ask him under what con- 
litions it could or would have been a Morris. ' Could have ' is 
lere used to concede that, although I happen to know it was 
.n Austin, your guess that it was a Morris was not a bad one. 
Jut ' could have ' also has a use which is more important for 
•ur purpose and in which, as I shall try to show, it is equiva- 
ent to 'would have . . . if . . .'. It refers to a tendency or 
apacity. Consider the following examples : 

(1) He could have read Emma in bed last night, though he 
ctually read Persuasion; but he could not have read Werther 
lecause he does not know German. 

(2) He could have played the Appassionata, though he 
ctually played the Moonlight; but he could not have played 
he Hanimerklavier, because it is too difficult for him. 

These are both statements, since they could be true or false ; 
nd to understand their logic we must see how they would be 
stablished or rebutted. Neither could be established or re- 
lutted in the way that ' He read Persuasion ' could, by observ- 
ttg what he actually did; and it is partly for this reason that 
i^e do not call them categorical. But, although they could not 
le directly verified or falsified by observation of what he did, 
his might be relevant evidence. It would be almost conclusive 
vidence in the first case, since it would be very odd if a man 
i^ho actually read Persuasion was incapable of reading Emma. 
)n the other hand, his having played the Moonlight is only 
/eak evidence that he could have played the Appassionata, 
ince the latter is more difficult and also because he might 
lever have learnt it. 

In each of these cases, in order to establish the ' could have ' 
tatement we should have to show (a) that he has performed 
asks of similar difficulty sufficiently often to preclude the 



possibility of a fluke, and (b) that nothing prevented him or 
this occasion. For example we should have tO establish thai 
there was a copy of Emma in the house. 

Statements about capacities, whether of the ' can ' or of th( 
'could have' kind, contextually imply unspecified conditions 
under which alone the person might succeed; and ' could have 
statements can be refuted either by showing that some neces- 
sary condition was absent (there was no copy of Emma) oi 
by showing that the capacity was absent. The first point coulc 
be established directly. How could the second be established 
In practice we do this either by appealing to past performance: 
or failures or by asking him to try to do it now. It is cleai 
that neither of these methods could be applied directly to th< 
occasion in question. We know that he did not read Emma 
and it is nonsense to ask him to try to have read Emma las 
night. And the very fact that evidence for or against 'coulc 
have' statements must be drawn from occasions other thai 
that to which they refer is enough to show that ' he could hav( 
acted otherwise' is not a straightforward categorical state- 
ment, at least in the type of case we have been considering 
Whether it is possible or necessary to interpret it categorically 
in moral cases is a point which I shall examine in the ner 

It might be argued that the sort of evidence by which ' coulc 
have' statements are supported or rebutted is never conclu- 
sive; and this is true. The argument used is an inductive one 
with a special type of conclusion. We might use an ordinan 
inductive argument to predict his future performance fron 
known past performances or in support of a statement abou' 
an unknoAvn past performance. But in this special case we knov 
that he did not do the thing in question, because we know tha 
he did something else; so we put our conclusion in the forn 
'he could have done X'. 

Whatever the evidence, it is always open to a sceptic to saj 
"I know he has always succeeded (failed) in the past; but h( 
might have failed (succeeded) on this occasion ". Now this sor 
of scepticism is not peculiar to 'could have' statements; it is 



one variety of general scepticism about induction. It is possible 
that if I had tried to add 15 and 16 last night (which I did not) 
I should have failed ; but it is also possible that if I tried now 
I should fail. Our use of 'could have' statements, like our use 
of predictions and generalizations, always ignores such refined 
scepticism; and it would be absurd to try to base either free- 
dom or responsibility on the logical possibility of such contin- 
gencies. In practice we ignore the sceptic unless he can 
produce reasons for his doubt, unless he can say why he be- 
lieves that a man who has always succeeded might have failed on 
just that occasion. If no such reason is forthcoming we always 
allow inductive evidence which establishes the existence of 
a general capacity to do something to establish also the 
statement that the man could have done it on a particular 
r^ccasion. Nor is this practice due to the fact that (the world 
being what it is) we are unfortunately unable to find better 
evidence and must fall back on probabilities. Our practice 
lies at the heart of the logic of 'can' and 'could have'. For the 
sceptic is, here as elsewhere, asking for the logically impos- 
sible ; he is asking us to adopt a criterion for deciding whether 
a man could have done something on a particular occasion 
which would make the words 'can' and 'could have' useless. 
What would be the result of accepting this suggestion? We 
should have to say that the only conclusive evidence that a 
man can do (could have done) X at time t is his actually 
doing (having done) X at time t. Thus the evidence that 
entitles us to say * He could have done X at time t ' would also 
entitle us to say 'He did X at time t\ and the 'could have' 
form would be otiose. 

Capacities are a sub-class of dispositions. To say that a man 
' can ' do something is not to say that he ever has or will ; there 
may be special reasons why the capacity is never exercised, 
For example that the occasion for exercising it has never 
irisen. A man might go through his whole life without ever 
idding 15 and 16; and we should not have to say that he 
wouldn't do this. Yet a man cannot be said to be able to do 
something if all the necessary conditions are fulfilled and he 



has a motive for doing it. It is logically odd to say " Smith can 
run a mile, has had several opportunities, is passionately fond 
of running, has no medical or other reasons for not doing so, 
but never has in fact done so". And, if it is true that this is 
logically odd, it follows that 'can' is equivalent to 'will . . . 
if . . .' and 'could have' to 'would have . . . if . . .'. To say that 
Smith could have read Emma last night is to say that he would 
have read it, if there had been a copy, if he had not been struck 
blind, etc., etc., and if he had wanted to read it more than he 
wanted to read anything else. Both the 'etc' and the last 
clause are important ; we cannot specify all the necessary con- 
ditions; and, granted that the conditions were present and that 
he could have read it, he might still not have read it because 
he did not want to. But if he did not want to do anything else 
more than he wanted to read Emma, he could not in these 
conditions be said to have chosen to do something else. He 
might have done something else, but not in the important 
sense of ' done ' which implies choosing. 


Libertarianism. Before considering why 'he could have acted 
otherwise', interpreted in this hypothetical way, is regarded 
as relevant to ascriptions of responsibility, it is necessary to 
examine the theory that, although the hypothetical interpre- 
tation is correct in most cases, in the special case of moral 
choice the phrase must be interpreted in a categorical way. It 
would indeed be remarkable if modal forms which are nor- 
mally used in a hypothetical way were used categorically in 
one type of case alone ; and I have already suggested that their 
logic is partly determined by the method that would be used 
to support or rebut statements which employ them. The thesis 
that ' he could have acted otherwise ' is categorical is equivalent 
to the thesis that it could be verified or falsified by direct 
observation of the situation to which it refers. 

It is essential to notice that the categorical interpretation is 



supposed to be necessary only in a very small, but very impor- 
:ant part of the whole range of human choice. And this too is 
•emarkable; for it implies that the words 'free' and 'choose' 
ire logically different in moral and in non-moral cases. There 
s a sense of 'free' to which I have already alluded in which 
t is contrasted with 'under compulsion'; and in this sense 
ictions are still free when they are completely determined by 
he agent's tastes and character. For to say that they are deter- 
nined in this way is not to say that he is a Pawn in the hands 
)f Fate or a Prisoner in the iron grip of Necessity. It is only 
o say that anyone who knew his tastes and character well 
;nough could predict what he will do. The fact that we can 
)redict with a high degree of probability hov/ Sir Winston 
Ilhurchill will vote at the next election does not imply that he 
loes not cast his vote freely. To be 'free' in this sense is to 
»e free to do what one wants to do, not to be able to act in 
pite of one's desires. 

According to the theory to be examined most of our volun- 
ary actions are 'free' only in this sense which implies no 
»reach in causal continuity. I choose what I choose because 
ay desires are what they are ; and they have been moulded by 
ountless influences from my birth or earlier. But, it is said, 
wral choices are free in a quite different sense, and one that 
3 incompatible with their being predictable. This unpredicta- 
•ility is an essential feature in the categorical interpreta- 
ion of 'he could have acted otherwise'; for, if anyone could 
tredict what I am going to do, I should not really be choosing 
tetween genuinely open alternatives, although I might think 

Professor Campbell puts the contrast in the following way: 
'Free will does not operate in those practical situations in 
srhich no conflict arises in the agent's mind between what he 
onceives to be his ' duty ' and what he feels to be his ' strongest 
lesire'. It does not operate here because there is just no 
iccasion for it to operate. There is no reason whatever why 
he agent should here even contemplate choosing any course 
(ther than that prescribed by his strongest desire. In all such 



situations, therefore, he naturally wills in accordance with his 
strongest desire. But his ' strongest desire ' is simply the speci- 
fic expression of that system of conative and emotive disposi- 
tions which we call his 'character'. In all such situations, 
therefore, whatever may be the case elsewhere, his will is in 
effect determined by his character as so far formed. . . ." 

. . . (On the other hand) "in the situation of moral conflict, 
I, as agent, have before my mind a course of action, X, which 
I believe to be my duty ; and also a course of action, Y, incom- 
patible with X, which I feel to be that which I most strongly 
desire. Y is, as it is sometimes expressed, 'in the line of least 
resistance ' for me - the course which I am aware that I should 
take, if I let my purely desiring nature operate without hin- 
drance. It is the course towards which I am aware that my 
character, as so far formed, naturally incHnes me. Now, as 
actually engaged in this situation, I find that I cannot help 
believing that I can rise to duty and choose X; the 'rising to 
duty' being affected by what is commonly called 'effort of 
will'. And I further find, if I ask m.yself just what it is I am 
believing when I believe that I ' can ' rise to duty, that I cannot 
help believing that it lies with me, here and now, quite abso- 
lutely, which of two genuinely open possibilities I adopt ; 
whether, that is, I make the effort of will and choose X or, 
on the other hand, let my desiring nature, my character as so 
far formed, 'have its way', and choose Y, the course in the line 
of least resistance."^ 

Now it is certainly true that many determinists have paid 
too little attention to the concept of 'trying' or 'making an 
effort ' ; but I think that there are certain difficulties in Professor 
Campbell's account of moral conflict and, in particular, in his 
attempt to construe 'I could have acted otherwise' in a cate- 
gorical way. The first point to which I wish to draw attention 
is the question of method. 

(i) Campbell insists that the question whether choice is 
'free' in a contra-causal sense must be settled by introspec- 
tion.2 But is this so ? To doubt the findings of his self-examina- 

7.. Mind, 1951, pp. 460-3. 2. Scepticism and Construction, p. 131. 



tion may seem impertinent; but the doubt is concerned, not 
with what he finds, but with the propriety of the language he 
uses to describe what he finds. The universal negative form 
of statement ('Nothing caused my decision', 'No one could- 
have predicted my decision') does not seem to be a proper 
vehicle for anything that one could be said to observe in self- 
examination. That I know introspectively what it is like to 
choose may be true; but I cannot be said to know introspec- 
tively that my choice was contra-causal or unpredictable; and 
this is the point at issue. He represents 'I can rise to duty' as 
a report of a mental event or, perhaps, a state of mind, not 
as a statement about a capacity, and 'I could have . . .' as a 
statement about a past state of mind or mental event. But, if " 
this is really so, it is at least surprising that, in this one con- 
text alone, we use the modal words 'can' and 'could have' for 
making categorical reports. The issue between determinists 
and libertarians is an issue about the way in which expressions 
such as 'choose', 'can', and 'alternative possibilities' are to be 
construed ; and this is surely an issue which is to be settled not 
by self-observation but by logical analysis. 

There are many other phrases in Campbell's account which 
give rise to the same doubts about the propriety of the intro- 
spective method. The phrase 'conative disposition' is em- 
bedded in a large and complex mass of psychological theory 
and its use implies the acceptance of this theory; so that one 
could hardly be said to know by introspection that one has a 
conative tendency to do something. And phrases such as 
'determined', 'contra-causal', and even 'desiring nature' take 
us beyond psychology into metaphysics. To say this is not to 
condemn the phrases; perhaps metaphysics is just what is 
needed here. But a metaphysician is not a reporter; he is an 
interpreter of what he ' sees ' ; and it is over the interpretation 
that the disputes arise. 

(2) A more obvious difficulty - and it is one of which 
libertarians are well aware - is that of distinguishing a 'free' 
action from a random event. The essence of Campbell's 
account is that the action should not be predictable from a 



knowledge of the agent's character. But, if this is so, can what 
he does be called Aw action at all? Is it not rather a lusus 
naturae, an Act of God or a miracle ? If a hardened criminal, 
bent on robbing the poor-box, suddenly and inexplicably fails 
to do so, we should not say that he chose to resist or deserves 
credit for resisting the temptation ; we should say, if we were 
religious, that he was the recipient of a sudden outpouring of 
Divine Grace or, if we were irreligious, that his ' actian ' was 
due to chance, which is another way of saying that it was inex- 
plicable. In either case we shovild refuse to use the active voice. 

The reply to this criticism is that we must distinguish /wde- 
terminism from ^e/f-determinism. Choice is a creative act of 
the 'self and is not only unconstrained by external forces but 
also unconstrained by desire or character. But the difficulty 
here is to construe 'self-determinism' in such a way that the 
'self can be distinguished from the 'character' without lapsing 
into indeterminism. 

If we could construe ' self-determined ' by analogy with other 
'self -compounds, such as self-adjusting, self-regulating, self- 
propelled, self-centred, self-controlled, and self-governing, 
there would be no difficulty. Some of these words apply to 
non-human objects, and they never imply that there is a part 
of the object called the 'self which adjusts, regulates, or con- 
trols the rest, though the object does have a special part with- 
out which it would not he self-adjusting, etc. I can point to 
the self-starter of a car, but not to the self that starts the car ; 
to say that a heating system is 'self- regulating' is to say that 
it maintains a constant temperature without anyone watching 
the dials and turning the knobs. Coming to the human scene, to 
say that a state is 'self-governing' is to say that its inhabi- 
tants make their own laws without foreign intervention; and 
to say that a man is 'self-centred' is to say, not that he is 
always thinking and talking about something called his 'self, 
but that he is always thinking and talking about his dinner, 
his golf-handicap, the virtues of his wife, and the prowess of 
his children. In each case there is a subject and an object; but 
the 'self is neither subject nor object. 



But if we construe ' self-determined ' in this way, it is clear 
that being self-determined implies only that a man acts freely 
in the ordinary sense of 'freely' which the libertarian rejects 
as inadequate in the special case of moral choice. There would 
be no incompatibility between an action's being 'self-deter- 
mined' and its being predictable or characteristic of the agent; 
for ' self-determined ' would mean ' determined by his motives 
and character ', as opposed to ' forced on him by circumstances 
or other people'. But the Hbertarian regards explanation in 
terms of character as incompatible with genuine freedom and 
must therefore draw a contrast between 'the self and 'the 
character'. But if ' self-determined ' is to mean 'determined by 
the self, it is necessary to give some account of what the 
'self is. And if the question whether an action was deter- 
mined by the 'self or not is to be relevant to the ascription 
of responsibility and the justice of adverse verdicts, we must 
be able to provide some criterion for deciding whether the 
self which determined the action is the same self that we are 
proposing to hold responsible or condemn. 

Now the problem of Personal Identity is admittedly a diffi- 
cult one and the danger of desert-island argument is particu- 
larly acute here, since Jekyll-and-Hyde cases that a layman 
would dismiss as flights of fancy have been known to occur. 
In fact we decide whether the man I met yesterday is the same 
that I met last year partly by seeing whether he looks the 
same, partly by observing an identity of characteristic beha- 
viour, and partly by discovering what he can remember. And 
if we are to avoid the rather crude course of defining ' same 
self in terms of the spatio-temporal continuity of bodily cells, 
it seems that we must define it in terms of character and 
mem.ory. But the libertarian's 'self is neither an empirical 
object nor displayed in characteristic action. 

(3) If it is necessary to decide whether or not a man could 
have acted otherwise before ascribing responsibility, it is 
necessary that we should have some criterion for deciding this ; 
and on the libertarian theory such a criterion is quite impos- 
sible. For, let us suppose that we know a great deal about his 



character and also that the temptation which he faced seems 
to be a fairly easy one for such a man to overcome. On the 
libertarian hypothesis this information will not be sufficient to 
enable us to conclude that he could have acted otherwise. If he 
in fact does the wrong thing, there are three alternative con- 
clusions that we might draw, (a) The action was not against 
his moral principles at all, so that no conflict between ' duty ' 
and 'inchnation' arose. This is what I have called 'wicked- 
ness'; (b) he knew it was wrong and could have resisted the 
temptation but did not (moral weakness); (c) he knew it 
was wrong but the temptation was too strong for him; he 
could not overcome it (addiction). Now it is essential to be 
able to distinguish case (b) from case (c), since (b) is a 
culpable state while (c) is not. By treating 'he could have 
acted otherwise' in a hypothetical way, the determinist thesis 
does provide us with a criterion for distinguishing between 
these cases; but the categorical interpretation cannot provide 
one, since no one, not even the man himself, could know 
whether he could have overcome the temptation or not. 

(4) The libertarian theory involves putting a very special 
construction on the principle that ' ought ' implies ' can ', which 
it is very doubtful whether it can bear. If we take this principle 
in a common-sense way it is undoubtedly true. It is no longer 
my duty to keep a promise, if I literally cannot do so. But 
when we say this we have in mind such possibilities as my 
being detained by the police or having a railway accident or 
the death of the promisee; and it is possible to discover 
empirically whether any of these exonerating conditions ob- 
tained. But if ' cannot ' is construed in such a way that it covers 
my being too dishonest a person or not making the necessary 
eflFort, it is no longer obvious that ' ought ' implies ' can '. These 
reasons for failure, so far from exonerating, are just what 
make a man culpable. 

(5) Even if it were possible to discover whether or not a 
man could have acted otherwise by attending to the actual 
occasion, as the categorical interpretation insists, why should 
this be held relevant to the question whether or not he is to 



blame? I shall try to explain this connexion in the next 
chapter; but on the libertarian hypothesis it will, I think, be 
necessary to fall back on insight into a relation of fittingness 
between freedom and culpability. 


The Concept of ' Trying \ It might be thought that the liber- 
tarian could discover a criterion for distinguishing culpable 
weakness of will from non-culpable addiction in the concept 
of ' trying '. For the addict fails, try as he may, while the weak- 
willed man fails because he does not try hard enough. The 
concept of ' trying ' is an important one for ethics since, what- 
ever may be the case in a court of law, the question of moral 
blameworthiness often turns, not on what the agent did, but 
on what he tried or did not try to do. Morally we blame 
people, not for failing to live up to a certain standard, but for 
not trying hard enough to do so; and this is because, while 
we do not believe that they could always succeed, we do be- 
lieve that they could always try. We must now see whether 
the introduction of this concept helps to save the categorical 

We all know what it feels like to make an effort. These 
feelings are phenom.ena or occurrences that we experience in 
the same sort of way that we experience aches, pains, qualms, 
and twinges. And, if "\ve take the introspective language of the 
libertarian seriously, it would seem that the question ' Did he 
try?' can be answered only by the man himself and that he 
answers it by observing whether or not one of these feelings 
occurred. The logical status of this question will be like that 
of 'Did it hurt?'. But on this view an eifort is not something 
that a man makes; it is something that happens to (or inside) 
him; and it would be highly unplausible to make the question 
of his responsibility turn on the occurrence or non-occurrence 
of such a feeling. If 'making an effort' is to be relevant to 



responsibility, it must be thought of as something which a 
man can choose to do or not to do. The substitution of the 
active for the passive voice is an important advance; unfor- 
tunately it is fatal to the categorical interpretation of ' he could 
have acted otherwise '. 

For 'trying' is now thought of as som.ething that a man 
can choose to do or not to do, and the difficulties encountered 
in construing 'he could have acted otherwise' will emerge 
again in construing ' he could have tried to act otherwise '. On 
the libertarian analysis, if a man fails to act rightly, we must 
say either that his failure is inexplicable or that it was due to 
circumstances beyond his control - in which cases he is 
blameless - or that it was due to his not having tried as hard 
as he could have tried. For what exonerates is not 'I tried', 
but ' I tried as hard as I could ' ; and, in order to distinguish 
the blameworthy man from the addict who literally couldn't 
help it because he tried as hard as he could, we must be in a 
position to answer the question ' Could he have tried harder 
than he did?'. But how can we answer this question? Ex 
hypothesi he did not try harder than he did; so that we must 
say either that his failure to try harder is inexplicable or that 
it was due to circumstances beyond his control - in which 
cases he is blameless - or that it was due to his not having 
tried to try as hard as he could have tried to try. 

But this is absurd. In the first place ' try to try ' is meaning- 
less; and, if this be doubted, we must push the analysis one 
stage further. In fact he did not try to try harder than he did. 
But can he be justly blamed for this ? Only if he could have 
tried to try harder. We must say either that his not having 
tried to try harder is inexplicable or that it was due to circum- 
stances beyond his control - in which cases he is blameless - 
or that he failed to try to try harder because he did not try 
to try to try harder . . . and so on. 

Libertarians sometimes speak in terms of our failure to 
make the best use of our stock of " will- energy " ; but this 
usage gives rise to the same infinite regress. If using will- 
energy is thought of as something that we do not choose to 



do, but which just happens to us, it would appear to be 
irrelevant to responsibility; but if it is something that we 
can choose to do or not to do, we must be able to distinguish 
the man whose failure to use sufficient will-energy was due 
to circumstances beyond his control from the man who failed 
(culpably) to use it because he did not try hard enough to 
use it. And this involves answering the question 'Had he 
sufficient second-order will-energy to enable him to make 
more use of his first-order will-energy ? ' 

On these lines there is clearly no way out of the wood. The 
attempt to discover one is, I think, due to two mistakes, (a) It 
is noticeable that, on Campbell's analysis, a man's desires and 
even his character are continually referred to as 'it'; desires 
are thought of as forces which, sometimes successfully and 
sometimes unsuccessfully, prod a man into doing what he 
ought not, and his "character as so far formed" is the sum of 
these forces. Thus I am said to be able to choose whether or 
not to " let my desiring nature, my character as so far formed, 
have its way". And this is to treat all cases of 'doing what I 
want to do ' on the model of the opium-addict, as the actions 
of a man who is a slave to his desire. 

And since Campbell uses 'desire' for every motive except 
the sense of duty, his treatment presupposes that I can choose 
whether to act from a certain motive or not; and this is not 
so. If I am both hungry and thirsty I can choose whether to 
have a meal or a drink; but I cannot choose whether to act 
from hunger or thirst, unless this strange phrase is used simply 
as a (very misleading) synonym for ' choosing whether to eat 
or to drink '. In the same way, if I have a certain sum of money, 
I can choose whether to pay a debt or give my aunt a Christ- 
mas present. If I choose the former, my motive -is conscien- 
tiousness ; if the latter, it is generosity. And we might, there- 
fore, say that I can choose whether to do the conscientious or 
the generous thing. But I cannot choose whether to act from 
conscientiousness or from generosity. What I do will depend 
on my character ; and this ' cannot choose ' is not a lamentable 
restriction on my freedom of action. For to say that my choice 



depends on my character is not to say that my character com- 
pels me to do what I do, but to say that the choice was charac- 
teristic of me. The creative 'self' that sits above the battle of 
motives and chooses between them seems to be a legacy of the 
theory that a man is not free when he does what he wants to 
do, since he is then the victim or slave of his desires ; and it 
is postulated to avoid the unplausible doctrine that all action 
is involuntary. 

(b) Campbell takes as a typical and, by implication, the 
only case of moral choice to which appraisals are relevant, that 
of a man who knows what he ought to do but is tempted to do 
something else. Now this, so far from being the only case, is 
not even the commonest or most important. For in the great 
majority of cases of moral difficulty what is difficult is not to 
decide to do what one knows one ought to do, but to decide 
what one ought to do. This sort of difficulty arises in three 
main types of case, (i) A humble and unimaginative person 
who accepts a customary code of morals without much ques- 
tion may find that two rules conflict; the voice of conscience 
is in this case ambiguous, (ii) A more self-confident, imagi- 
native, and reflective person may v/onder v/hether he ought, 
in the case before him, to do what the customar}^ rule enjoins. 
He knows very well what the rule enjoins; but what prompts 
him to depart from it is not " part of his desiring nature", but 
a suspicion that the rule is one that, in this particular case, he 
ought not to follow, (iii) A man of fixed moral principles 
(whether or not they are those customarily adopted) may 
find himself in a radically new situation that is not catered for 
in his code. What is he to do? It is here, if anywhere, that 
the idea of an unpredictable ' creative ' choice seems to make 
sense. He takes a leap in the dark, but just because it is a leap 
in the dark I doubt if we should be inclined to blame him if he 
leapt in what turned out to be the wrong direction. 

Men who belong to a generation for whom the questioning 
of accepted principles has been no mere academic exercise and 
who have found themselves faced with momentous choices in 
situations not covered by their traditional rules will be less 



I - ' 

' fikely than their fathers perhaps were to suppose that the only- 
sort of moral difficulty is that of resisting temptation. 

If, in the first two of these three cases, a man decided that 
he ought to do something and did it, he might still be held 

I to blam.e. For reasons given in chapter 17 conscientiousness 
is so valuable a motive that we should be chary of blaming a 
man who did what he honestly thought he ought to do, how- 
ever misguided we thought him. But we should not necessarily 
excuse him, which we should have to do if all wrong- doing 
were failure to resist temptation. Integrity is not the only 
moral virtue, any more than it is the only virtue in an artist; 

■ and the belief that it is is one of the more regrettable conse- 

1 quences of the Romantic Movement. We blame people, not 
only for failing to live up to their moral principles, but also 
for having bad moral principles ; and I shall examine the logic 

I of this type of blame in the next chapter. 

Perhaps the most crucial objection to the libertarian thesis 
lies in the sharp discontinuity which it presupposes between 
moral and non-moral choice and between moral and non-moral 
appraisal. It is not enough to admit that we can, within broad 
limits, predict what a man of known habits, tastes, and in- 
terests will do and to insist that our powers of prediction only 
break down in the small, but important area of moral choice. 
For it is not the extent of the area open to prediction that is at 

It is true that we can, within broad limits, predict what a 
man will choose from a menu, whether he will make a century 
today, or finish his crossvvord puzzle ; but we can also predict, 
again vv'ithin broad limits only, whether or not he will resist 
the temptation to run away or to cheat at cards. Our reliance 
on the integrity of a bank clerk is not different from our 
reliance on his accuracy. In neither case do we believe that he 
* must ' or ' is compelled to ' be honest or accurate ; and what is 
paradoxical is not so much the libertarian's defence of moral 
freedom as his willingness to accept mechanical determinism 
as an explanation of non-moral action. For the rigid distinc- 
tion between 'formed character' (where determinism reigns) 



and 'creative choice' (which is in principle unpredictable) it 
would be better to substitute a conception of continual modi- 
fication of character in both its moral and its non-moral 
aspects. This not only does justice to the fact that we use 
both choosing and appraising language in the same way in 
moral and non-moral contexts, but it is closer to the facts. A 
man can grow more or less conscientious as time goes on, 
just as he can become better at tennis or more fond of Mozart. 

Note to Section [2] 

J. L. Austin has shown that the analysis of the concept of 
ability in Section [2] of this chapter is unsatisfactory, and in 
particular that the argument in the last paragraph is wrong. 
He has made it clear, first, that ' could have ' is, in the relevant 
cases, not a subjunctive indicating a suppressed conditional, 
but the past indicative of ' can '. ' He could have read Emma ' 
means ' he was able . . .', not ' he would have been able . . . if . . .'. 
Secondly it can be shown by the methods of formal logic 
that we cannot argue from the logical oddness of ' Smith can 
run a mile . . . but never has in fact done so ' to the conclusion 
that ' can ' means ' will . . . if . . .' (p. 278). 

Nevertheless I remain convinced that statements made 
with 'can' are not straightforward categoricals asserting the 
existence of a state of affairs, and that the concept of being 
able to do something, which evidently lies at the heart of the 
free-will problem, must be analysed in terms of actual per- 
formances. Readers who wish to navigate further into these 
treacherous waters should consult J. L. Austin : Philosophical 
Papers, pp. 153-80, and my reply in Theoria, vol. xxvi, no. 2, 
January i960. 



Freedom and Responsibility (2) 


In the last chapter I tried to show that ' could have ' sentences 
in non-moral contexts can be analysed in terms of ' would have 
. . . if . . .' ; and we must now see whether the application of 
this analysis to moral cases is consistent with our ordinary use 
of moral language. 

The first question to be considered is the question what 
sorts of if-clauses are in fact allowed to excuse a man from 
blame. Clearly ' I could not have kept my promise because I 
was kidnapped' will exculpate me while 'I could not have 
kept my promise because I am by nature a person who takes 
promises very lightly' will not. Translated into the hypo- 
thetical form, these become respectively 'I would have kept 
my promise if I had not been kidnapped ' and ' I would have 
kept my promise if I had been a more conscientious person '. 
Again it is clear that the first exculpates while the second does 
not. The philosophical difficulties, however, are to decide just 
why some 'would . . . ifs' excuse while others do not and to 
provide a criterion for distinguishing the exculpating from the 
non-exculpating cases. Forcible seizure exculpates; but do 
threats or psychological compulsion ? And if, as some suggest, 
desires are internal forces which operate on the will, do they 
exculpate in the way in which external forces do ? The prob- 
lem o\ free will is puzzling just because it seems impossible, 
without indulging in sheer dogmatism, to know just where to 
stop treating desires as ' compelling forces '. 

Now before tackling this difficulty it will be prudent to 
examine what goes on in a place where questions of respon- 
sibility are settled every day and have been settled daily for 
hundreds of years, namely a court of law. Lawyers have 



evolved a terminology of remarkable flexibility, refinement, 
and precision and, although there may be a difference between 
moral and legal verdicts, it would be strange if the logic of 
lawyers' talk about responsibility were very different from our 
ordinary moral talk. 

To establish a verdict of * guilty ' in a criminal case it is neces- 
sary to establish that the accused did that which is forbidden 
by the law or, in technical language, committed the actus reus, 
and also that he had what is called mens rea. This last phrase 
is sometimes translated 'guilty mind' and in many modern 
textbooks of jurisprudence it is supposed to consist of tv/o ele- 
ments, (a) foresight of the consequences and (b) voluntari- 
ness. But, whatever the textbooks may say, in actual practice 
lawyers never look for a positive ingredient called volition 
or voluntariness. A man is held to have mens rea, and there- 
fore to be guilty, if the actus reus is proved, unless there are 
certain specific conditions which preclude a verdict of guilty. 
"What is meant by the mental element in criminal Hability 
{mens rea) is only to be understood by considering certain 
defences or exceptions, such as Mistake of Fact, Accident, 
Coercion, Duress, Provocation, Insanity, Infancy."^ The list 
of pleas that can be put up to rebut criminal liabihty is dif- 
ferent in different cases ; but in the case of any given offence 
there is a restricted list of definite pleas which will preclude 
a verdict of guilty. 

This is not to say that the burden of proof passes to the 
defence. In some cases, such as murder, it is necessary for 
the prosecution to show that certain circumstances were not 
present which would, if present, defeat the accusation. The es- 
sential point is that the concept of a ' voluntary action ' is a nega- 
tive, not a positive one. To say that a man acted voluntarily 
is in effect to say that he did something when he was not 
in one of the conditions specified in the list of conditions which 
preclude responsibility. The list of pleas is not exhaustive; we 

I. Professor H. L. A. Hart: Proceedings of the Aristotelian Society, 
1948-9. Aristotle in effect defines 'the voluntary' in the same negative 
way as what is done not under compulsion and not through ignorance. 



could, if we wished, add to it; and in making moral judge- 
ments we do so. For example we sometimes allow the fact that 
a man acted impulsively to exonerate him morally or at least 
to mitigate his offence in a case in which the law would not 
allow this. But it remains true that, in deciding whether an 
action was voluntary or not, we do not look for a positive 
ingredient but rather for considerations that would preclude 
its being voluntary and thereby exonerate the agent. In moral 
cases the most important types of plea that a man can put for- 
ward are (a) that he was the victim of certain sorts of ignor- 
ance, and (b) that he was the victim of certain sorts of 

Ignorance. A may be ignorant of many elements in the 
situation in which he acts. For example he may not Icnow that 
it was a policeman who told him to stop, that the stuff he put 
in the soup was arsenic, that the money he took was not his 
ov/n. In such cases he would be blamed only if it was thought 
that he ought to have known or taken the trouble to find out. 
And his vicious trait of character was not contumacy or cal- 
lousness or greed or disregard for any moral principle, but 
carelessness ; and carelessness can amount to a vice. Fire-arms 
are so notoriously dangerous that the excuse *I didn't Icnow 
it was loaded' will not do. The reason why he is blamed for 
carelessness and not for the specific vice for which he would 
have been blamed if he had done any of these things inten- 
tionally is that, although he intended to do what he did, he did 
not intend to break a moral rule. He intended to take the 
money, but not to steal. His action was not, therefore, a mani- 
festation of the particular vice that the actions of thieves mani- 
fest. Ignorance of fact excuses or reduces the seriousness of an 
offence ; but there is one type of ignorance that never excuses ; 
and that is, in legal contexts, ignorance of the law and, in 
moral contexts, ignorance of right and wrong. 



Now why should ignorance of fact excuse while ignorance 
of rules does not? Why should a man who takes someone else's 
money, thinking it to be his own, be guiltless of anything 
(except possibly carelessness), while a man who takes it, 
knowing it not to be his own but because he sees nothing 
wrong in taking other people's money, be held guilty and 
therefore blameworthy? We are not here concerned with the 
question why some types of action should be stigmatized as 
'wrong', but solely with the question why ignorance of what is 
wrong should not be held to exculpate. 

The reason is that while the man who thought the money 
was his own did not intend to act on the maxim ' It is permitted 
to take other people's money ', the thief does act on this maxim. 
If a man does something because he does not think it wrong he 
cannot plead that he did not choose to do it, and it is for 
choosing to do what is in fact wrong, whether he knows it or 
not, that a man is blamed. The situation is exactly analogous 
to that in which some non-moral capacity is concerned. 'I 
would have solved the problem, if I had known all the data' 
would, if substantiated, allow me to get full marks. But ' I 
would have solved the problem if I had known more mathe- 
matics' would not. Since competence at mathematics is not a 
moral trait of character, men are not blamed for lack of it ; but 
they are given low marks and denied prizes. 


Compulsion. So long as ' compulsion ' is used in the literal sense 
it is not difficult to see why it should be held to exonerate. If a 
man is compelled to do something, he does not choose to do 
it and his action is not a manifestation of his moral character 
or principles. Now, since the purpose of blame and punish- 
ment is to change a man's character and principles, neither 
blame nor punishment is called for in such a case. It would be 
unjust to punish him since the rules for punishing lay down 



that a man who acts under compulsion is not to be punished; 
and the rules lay this down because, with due allowance for 
superstition and stupidity, we do not have pointless rules. 
Once more we must be careful to avoid the mistake of say- 
ing that the justice of a sentence turns on the question 
whether the accused is likely to be reformed by it. What is 
at issue here is not our reason for exonerating this accused, 
but our reason for making a general exception in the case 
of men whose actions are not expressions of their moral 
character. Physical compulsion is an obvious case where this 
is so. 

But what if the source of compulsion is within the man him- 
self? It is not an accident that we use 'compulsion' in a psy- 
chological way and exonerate compulsives. There are two 
questions that are relevant here. In the first place we ask 
whether the man could have resisted the ' compulsion ' ; and we 
decide this in the way that we decide all ' could have ' questions. 
We look for evidence of his past behaviour in this, and also in 
related matters ; for the behaviour of the compulsive is usually 
odd in matters unconnected with his special compulsion; and 
we compare his case with other known cases. Once the capacity 
to resist the compulsion is established beyond reasonable 
doubt we do not allow unsupported sceptical doubts about his 
capacity to resist it in a particular case to rebut the conclusion 
that he could have helped it. And we do not allow this because 
there is no way of establishing or refuting the existence of a 
capacity except by appeal to general evidence. If the capacity 
has been established and all the necessary conditions were 
present, we would not say that, in this case, he was the victim 
of a compulsion. Indeed a ' compulsion ' is not something that 
could be said to operate in a particular case only; for to say 
that a man has a psychological compulsion is to say something 
about his behaviour over a long period. A compulsion is more 
like a chronic disorder than like a cold ; and it is still less like a 

It is also relevant to raise the question whether he had any 
motive for doing what he did. Part of the diiference between 



a kleptomaniac and a thief lies in the fact that the former has 
no motive for what he does ; and he escapes blame because the 
point of blame is to strengthen some motives and weaken 
others. We are sometimes inclined to take the psychologists' 
talk about compulsions too seriously. We think that a man is 
excused because he has a ' compulsion ', as if the compulsion 
could be pointed to in the way that an external object which 
pushed him could be pointed to. But compulsions are not 
objects inside us; and we use the word 'compulsion', not 
because we have isolated and identified the object which caused 
him to do what he did, but because we want to excuse him in 
the same sort of way that we excuse someone who is literally 
pushed; and we want to excuse him for the same sort of| 
reason. We Icnow that it will do no good to punish him. | 

Desires. A man might plead that he would have acted other-' 
wise if he had not had a strong desire to do what he did ; but 
the desire was so strong that, as things were, he could not 
have acted otherwise. Would this plea be allowed to exonerate' 
him? In some cases it would; for there are, as we have seen, 
cases of addiction in which we allow that a man is not to 
blame since his craving was too strong for him. But in most! 
cases it would be considered frivolous to say *I would have* 
done the right thing if I hadn't wanted to do the wrong thing';? 
for it is just for this that men are blamed. \ 

To distinguish an overwhelming desire from one that the 
agent could have resisted is not always easy; but the criterion 
that we in fact use for making the distinction is not difficult to 
understand. We know from experience that most men can bej 
trained to curb some desires, but not others; and we assume' 
that what is true in most cases is true in a given case unless 
special reasons are given for doubting this. Now it might seem 
that, although this evidence enables us to predict that we shall 
be able to train the man to curb his desire in future, it sheds 
no light on the question whether he could have curbed it on 
the occasion in question. I shall say more about this question | 
of moral training later ; here I only wish to point out that we 
have no criterion for deciding whether a man could have 



resisted a desire on a given occasion other than general evidence 
of his capacity and the capacity of others like him. We do not, 
because v/e cannot, try to answer this question as if it referred 
solely to the given occasion; we treat it as a question about a 

Character. Finally a man might plead that he could not help 
doing what he did because that's the sort of man he is. He 
would not have done it if he had been more honest or less 
cowardly or less mean and so on. This sort of plea is para- 
doxical in the same sort of way that the plea of ignorance of 
moral rules and the plpa that he did it because he wanted to 
are paradoxical. And all three paradoxes stem from the same 
source, the uncritical extension of ' ought implies can ' and of 
the exculpatory force of 'he could not have acted otherwise* 
to cases which they will not cover. We know that these pleas 
are not in fact accepted; the puzzle is to see why. 

The plea ' I could not help it because I am that sort of per- 
son ' might be backed up by an explanation of how I came to 
be that sort of person. Just as the discovery of a compelling 
cause exonerates, so, it might be argued, to reveal the causes 
of my character being what it is is to show that I could not 
help being what I am and thus to exonerate me. But this 
argument is fallacious. In the first place to discover the cause 
of something is not to prove that it is inevitable. On the con- 
trary the discovery of the cause of a disease is often the first 
step towards preventing it. 

Now it is logically impossible to prevent something happen- 
ing if we know the cause of it, since it could not have a cause 
unless it occurred and therefore it was not prevented. So 
when we talk of preventing diseases or accidents we are 
not talking about preventing cases which have occurred but 
about ensuring that there are no future cases. Similarly, if I 
know how Jones came to be a dishonest man I cannot prevent 
him from being dishonest now; but it may be possible to 
prevent others from becoming dishonest and to cure Jones of 
his dishonesty. 

Secondly, the discovery of a cause of something has no 



necessary bearing on a verdict about that thing. We know that 
a man has come to be what he is because of three main types 
of cause, heredity, education, and his own past actions. These 
three factors are not independent of each other and it is not 
the business of a philosopher to say exactly what is the effect 
of each or which is the most important for moral training. The 
question ' Granted that we want people to be better and that 
we have fairly clear ideas about what "being better" means, 
should we try to breed a superior race or pay more attention 
to education?' is not a philosophical question. But it is the 
business of a philosopher to show in what ways these ' causes ' 
are related to responsibility. 

Now these three factors also play a part in situations in 
which non-moral verdicts are given. Leopold Mozart was a 
competent musician; his son Wolfgang was given a good 
musical education and practised his art assiduously. Each of 
these facts helps to explain how he was able to compose and 
play so well. There is plenty of evidence that musical ability 
runs in families and still more of the effects of teaching and 
practice. But, having learnt these facts, we do not have the 
slightest tendency to say that, because Mozart's abilities were 
' due ' to heredity, teaching, and practice, his compositions were 
not ' really ' his own, or to abate one jot of our admiration. In 
the same way, however a man came by his moral principles, 
they are still his moral principles and he is praised or blamed 
for them. The plea that, being what he is he cannot help 
doing what he does, will no more save the wicked man than 
it will save the bad pianist or actor who has the rashness to 
expose his incompetence in public. Nor is he saved by being 
able to explain how he has come to be what he is. 

Hereditary tendencies are not causes and do not compel, 
although a man may inherit a tendency to some form of 
psychological compulsion. In general to say that a man has a 
tendency to do something is to say that he usually does it ; and 
to add that the tendency is hereditary is to say that liis father 
also used to do the same sort of thing; and neither of these 
facts has any tendency to exculpate. 



The belief that heredity or a bad upbringing excuse a man's 
jresent character is partly due to the false belief that to explain 
something is to assign an antecedent cause to it and that, to be 
/^oluntary, an action must be uncaused. But there is also 
I good reason for this belief. In fact we do sometimes allow 
;hese factors to exculpate; and if the question of explanation 
Nzs as irrelevant to the question of responsibility as I have 
suggested it would be hard to understand why we do this, 
^hy do we tend to deal less harshly with juvenile delinquents 
vho come from bad homes than with those who have had 
:very chance? The question is not one of justice, since it 
s not a question whether Jones ought to be punished, but 
vhether the law should lay down that people whose bad 
;haracters are due to certain causes should be punished. We 
nust therefore ask what is our reason for differentiating 
)etween two boys whose characters and actions are the same 
)ut who come respectively from bad and good homes. And the 
eason is that in the first case we have not had a chance to see 
vhat kindness and a good education could do, while in the 
econd we know that they have failed. Since punishment 
nvolves the infliction of pain and since it is a moral rule that 
mnecessary pain should not be inflicted, there is a general 
)resumption that people should not be punished if the same 
!nd could be achieved without the infliction of pain. This 
;onsideration is, of course, irrelevant to the question whether 
ones should be punished; but it is highly relevant to the 
question whether a distinction should be made between those 
vhose characters have come to be what they are because of a 
)ad education and those whose characters are bad in spite of a 
food one. 

But suppose a man should plead that he cannot now help 
loing what he does because his character was formed by his 
)wn earlier actions? This also will not excuse him. The logic 
)f this plea is that he did X because he was, at the time, the 
sort of man to do X and that he became this sort of man 
?ecause he did Y and Z in the past. But if he cannot be blamed 
'or doing X now, can he be blamed for having done Y and Z in 



the past ? It would seem that he cannot, for he will exculpate 1 
himself in exactly the same way. 

Once again the argument presupposes that if his present 
character can be explained in terms of what happened in the , 
past he necessarily escapes blame. The assumption is that a 
man's actions form a causal chain in which each necessitates 
the next. Now, if we suppose that, to be free, an action must 
be uncaused, either we shall find a genuinely uncaused action i 
at the beginning of the chain or we shall not. If we do not, ; 
then no action is culpable ; and if we do, then we must suppose 
that, while most of our actions are caused and therefore blame- * 
less, there was in the past some one uncaused action for which 
alone a man can be held responsible. This theory has in fact 
been held, although even in the history of philosophy it v/ould 
be hard to find another so bizarre. The objections to it are ' 
clear. In the first place we praise and blame people for what • 
they do now, not for what they might have done as babies; 4 
and secondly this hypothetical infantile action could hardly be 1 
said to be an action of the agent at all, since it is ex hypothesi \ 
inexplicable in terms of his character. 

The conclusion of the foregoing argument is that ' He could 1 
not have acted otherwise' does not always exculpate and, in 
particular, that it does not exculpate if the reason which is 
adduced to explain just why he could not have acted otherwise 
is that he was a man of a certain moral character. We have 
seen that 'He could have acted otherwise' is to be construed 
as 'He would have acted otherwise, if . . ,' and we have 
seen which types of 'if are not allowed to exculpate. We 
must now see why they are not. 


What is moral character? The key to the logical relationships 
between the five types of judgement seems to lie in the judge- 
ment of moral character (5). For (2) is thought to be a neces- 



sary condition of (i), (3), and (4) only because we exclude 
those cases of incapacity to act otherwise in which the incapa- 
city lies in the moral character. If it is due to an external force 
or to a ' compulsion ' (which we talk of as if it were an external 
force), or to some non-moral defect, the incapacity to act 
otherwise is allowed to excuse; but not if it is due to a moral 
defect. And it is now necessary to provide some criterion for 
deciding what a moral defect is. 

Moral traits of character are tendencies or dispositions to 
behave in certain ways. How are they to be distinguished from 
other tendencies ? If any tendency were to count as ' moral ' we 
should have to say that conformity to physical laws was a 
universal trait of human character and that susceptibility to 
colds was part of the moral character of a particular man. 

The first and most obvious limitation lies in the fact that the 
names of virtues and vices are not purely descriptive words. 
They are terms of praise and blame used to express approval 
and disapproval and to influence the conduct of the person 
whose character is appraised and also of others. These three 
functions are tied together in a way that should by now be 
familiar. Appraising, praising, and blaming are things that men 
do and can only be understood on the assumption that they 
do them for a purpose and use means adapted to their purpose. 
The logic of virtue- and vice-words is tailor-made to fit the 
purposes and conditions of their use. 

Men would not employ a special form of speech for changing 
the character and conduct of others unless they had a pro- 
attitude towards those changes ; so that the first limitation that 
can be put on ' moral character ' is that traits of character are 
tendencies to do things that arouse approval or disapproval. 
But moral verdicts do not just express the attitudes of the 
speaker; they are couched in impersonal language and imply 
accepted standards because the traits of character that a given 
man wants to strengthen or inhibit in others are usually those 
that other men also want to strengthen and inhibit. The imper- 
sonal language of morals implies a rough community of pro- 
and con-attitudes. Moreover men would not have adopted the 



moral language they have unless it was likely to achieve its 
purpose ; and its purpose is achieved because most men dislike 
disapprobation. The power of moral language is greatly en- 
hanced by the very facts which make impersonal moral 
language possible. No one likes to be universally condemned 
and most men are willing to take considerable pains to avoid 

But this limitation is not enough. There are many things 
for which men are applauded and condemned which do not 
count as parts of their moral character. A great musician, 
mathematician, actor, or athlete is applauded and rewarded 
for what he does and his ability may be called a ' virtue ', but not 
a moral virtue. Conversely, if a man fails to save a life because 
he cannot swim, we may regret his incapacity and urge him 
to learn, but his incapacity is not called a vice. 

A man may fail to achieve some worthy object because he 
is physically or intellectually incompetent, too weak or too 
stupid. But he may also fail because he is too cowardly or too 
dishonest or has too little regard for the welfare of others. 
Why do we call the first set of traits ' non-moral ' and never 
condemn them, while the second are called ' moral ' and con- 
demned ? It is clear that it will not help to say that we intuit 
a non-natural relation of fittingness which holds between 
blameworthiness and dishonesty or meanness but not between 
blameworthiness and physical weakness or stupidity. For this 
is only to say that the former traits deserve blame while the 
latter do not and that we cannot understand why. 

To discover why we draw the line in the way that we do we 
must first ask exactly where we draw it; and all that is neces- 
sary for this purpose is to construct two lists, the one of moral 
traits, the other of non-moral. Cowardice, avarice, cruelty, 
selfishness, idleness would go into the first list; clumsiness, 
physical weakness, stupidity, and anaemia into the second. The 
second list will, of course, contain items of many different 
sorts, since we are interested, not in the way in which non- 
moral characteristics differ from each other, but in the dis- 
tinction between moral and non-moral. 



If we construct these lists we shall find that the items in 
St I have two properties in common which the items in list 2 
o not have, (a) We believe that if a man's action can be ex- 
lained by reference to a list i characteristic, he could have 
cted otherwise. And it would appear at first sight that this is 
le crucial feature which distinguishes moral from non-moral 
haracteristics. Why does a schoolmaster punish a lazy boy 
ut not a stupid one for equally bad work if not because he 
elieves that the lazy boy could have done better while the 
tupid boy could not ? But why does the schoolmaster believe 
lis ? In fact he appeals to the evidence of past performance. 
)n the libertarian view this would scarcely be relevant, since 
le boy might not have been lazy in the past but was lazy at 
1st that moment. And perhaps his momentary laziness was 
o more under his control than the stupid boy's stupidity ? An 
Qalysis on these lines could hardly fail to lead to the para- 
oxical conclusion that no one has any reason whatever for 
scribing responsibility. And even if it were possible to answer 
le question whether he could have acted otherwise, we should 
e left with the question why this is considered relevant to the 
ropriety of holding him responsible. 

Moreover it would be circular to make the phrase ' he could 
ave acted otherwise' the distinguishing criterion of moral 
haracteristics; since, as we have seen, it is necessary to make 
se of the distinction between actions explained by reference 
3 moral, and actions explained by reference to non-moral 
haracteristics in order to elucidate the phrase ' he could have 
cted otherwise'. 

(b) There is, however, another element which all the 
haracteristics in list i have and those in list 2 do not. It is 
n empirical fact that list i characteristics can be strengthened 
r weakened by the fear of punishment or of an adverse 
erdict or the hope of a favourable verdict. And when we re- 
lember that the purpose of moral verdicts and of punishment is 
5 strengthen or weaken certain traits of character it is not 
ifficult to see that this feature, so far from being synthetically 
onnected with the notion of a ' moral ' characteristic, a virtue 



or a vice, is just what constitutes it. What traits of character 
can be strengthened or vi^eakened in this way is a matter of 
empirical fact. Knives can be sharpened, engines decarbonized, i 
fields fertilized, and dogs trained to do tricks. And men also , 
can be trained, within certain limits, to behave in some ways 
and not in others. Pleasure and pain, reward and punishment 
are the rudders by which human conduct is steered, the means , 
by which moral character is moulded; and 'moral' character 
is just that set of dispositions that can be moulded by these , 
means. Moral approval and disapproval play the same role. , 
It is not just an accident that they please and hurt and that 
they are used only in cases in which something is to be gained 
by pleasing or hurting. 

We might therefore say that moral traits of character are 
just those traits that are known to be amenable to praise or 
blame; and this would explain why we punish idle boys but 
not stupid ones, thieves but not kleptomaniacs, the sane but 
not the insane. This is not to say that amenability to praise 
and blame is what justifies either of these in a particular case; 
that, as we have seen, is a question to be decided by reference 
to the rules. But a breach of a moral rule is only considered 
to be culpable when it is attributable to the agent's character, 
his vice or moral weakness; and our theory is intended to 
explain just what is included in and what excluded from ' moral 
character' and to explain why this distinction should be con- 
sidered relevant to responsibility. 

According to this explanation there is no need to postulate 
any special insight into necessary connexions between the five 
moral judgements with which we started ; for the whole weight 
of the analysis is now seen to rest on the proposition that 
people only do those things which are either objects of a direct 
pro-attitude (i.e. that they want to do or enjoy doing for their 
own sake) or are believed to produce results towards which 
they have pro-attitudes. It is absurd to ask why a man who 
thinks that praise and blame will alter certain dispositions 
which he wishes to alter should praise and blame them. For 
this is a special case of the question 'Why do people adopt 



means that they believe to be the best means of achieving 
:heir ends ? ' ; and this is an absurd question in a way in which 
Why does a man deserve blame only if he acted voluntarily 
md has broken a moral rule ? ' is not. 

Nevertheless this way of tracing the connexions between 
pro-attitudes, moral rules, verdicts on character, and ascrip- 
:ions of responsibility is obviously too simple and schematic. 
[t is more like an account of the way in which moral lan- 
guage would be used by people who knew all the facts and 
thoroughly understood what they were doing than like a 
description of the way in which moral language is actually 
used. In practice these connexions are much looser than the 
theory suggests ; and there are two reasons for this. In the first 
place there is the inveterate conservatism of moral language. 
Even when it is known that a certain type of conduct, for 
5xample homosexuality, is not amenable to penal sanctions or 
moral disapproval, it is difficult to persuade people that it is not 
morally wrong. 

The second reason is more respectable. We are still very 
ignorant of the empirical facts of human nature, and this 
Ignorance both makes it wise for us to make moral judge- 
ments in accordance with a more or less rigid system of 
rules and also infects the logic of moral language. Our moral 
verdicts do not, therefore, always imply that the person 
condenmed has in fact done something ' bad ' or ' undesirable ' 
in a non-moral sense. An act of cowardice or dishonesty 
might, by chance, be attended with the happiest consequences; 
but it would still be blamed. But this fact does not involve any 
major modification in the theory that bad traits of character 
are those which (a) tend to bring about undesirable results 
in most cases and (b) are alterable by praise and blame. For, 
in deciding whether a trait of character is vicious or not, 
we consider its effects in the majority of cases. We do not 
want to reinforce a tendency to behave in a certain way 
just because it turns out, on rare occasions, to be beneficial. 
And, in making a moral judgement, we do not consider the 
actual consequences of the action concerned. Nor do we even 

L 305 


need to consider the consequences that such actions usually 
have. A man has broken faith or been cowardly or mean; we 
condemn him forthwith without considering why such actions 
are condemned. The fact that deceitful, cowardly, and mean 
actions are, by and large, harmful is relevant, not to the 
questions : ' Has Jones done wrong ? Is he a bad man ? Does he 
deserve to be blamed?', but to the question 'Why are deceit- 
fulness, cowardice, and meanness called "vices" and con- 
demned ? ' 

This theory enables us to understand why it is not only 
moral weakness that is blamed, but also wickedness; and it 
also enables us to distinguish between moral weakness and 
addiction in a way that the libertarian theory could not. A 
wicked character can be improved by moral censure and 
punishment; and if we really thought that a man was so bad 
as to be irremediable we should, I think, cease to blame him, 
though we might impose restraints on him as we would on a 
mad dog. Moral weakness is considered to be a less culpable 
state, since the morally weak man has moral principles 
which are good enough, but fails to live up to them. He is 
therefore more likely to be improved by encouragement than 
the wicked man is. What he needs is the confidence which 
comes from knowing that others are on the side of his prin- 
ciples. But both he and the wicked man differ from the addict 
or compulsive in that the latter will respond neither to threats 
nor to encouragement. 


Moral Principles. Traits of character, then, are dispositions to 
do things of which a spectator (including the agent himself) 
approves or disapproves and which can be, if not implanted or 
wholly eradicated, at least strengthened or weakened by 
favourable and adverse verdicts. But they are dispositions to 
do things, in the active sense of 'do', dispositions to choose 
certain courses of action. It is not, therefore, an accident that 



the ■ names of virtues and vices, such as ' generosity ' and 
'avarice', are motive-words which necessarily imply a pro- 
attitude towards doing the things called 'generous' or 
'greedy' for their own sake. And since moral principles 
are also dispositions to choose, they also must be classed 
as 'pro-attitudes'. How do they differ from other pro- 
attitudes ? 

(a) In the first place a pro-attitude does not count as a moral 
principle unless it is a relatively dominant one and concerned 
with an important matter. However regularly I choose to 
drink coffee for breakfast no one would call this disposition 
to choose one of my moral principles. To act on principle is 
consistently to pursue a policy of doing certain sorts of things 
for their own sake; and for this reason 'acting on principle' 
must be sharply distinguished from 'acting from a sense of 
duty', although we shall see later that the two are connected. 
The reason for distinguishing them is that to act from a sense 
of duty is consistently to pursue a policy of obeying certain 
rules for the sake of obeying those rules; it is therefore a 
special case of acting on principle. ' Acting on principle ' can- 
not, therefore, be identified with either the ' sense of duty ' or 
the ' impulses ' which, according to some philosophers, are the 
only types of motive. It is distinguished from ' acting on im- 
pulse ' by regularity and consistency and from ' acting from the 
sense of duty ' by the fact that the man who acts on principle 
does what he does for its own sake. 

Now since a moral principle is a disposition to choose, a 
man cannot be said to have a certain moral principle if he 
regularly breaks it, and we discover what a man's moral prin- 
ciples are mainly by seeing how he in fact conducts himself. 
But this is not the only test. A man's moral principles are 
'dominant' in the sense that he would not allow them to be 
over-ridden by any pro-attitude other than another moral 
principle. Thus a man may belong to many organizations and 
be allowed by the laws of his country to do something that he 
is not allowed to do by the rules of his trade union, profession, 
or church. When a conflict of principles or loyalties arises he 



may wonder what he ought to do ; but it is part of the force 
of the phrase 'moral principle' that he cannot (logically) 
wonder what he ought to do if there is a moral principle on one 
side and not on the other. If I regard something as immoral, 
then, however trivial it may be and however great may be the 
non-moral advantages of doing it, I cannot debate with myself 
whether I ought to do it ; and we discover what our own moral 
principles are very often by putting just this sort of question to 

A similar limitation in the use of the phrase ' moral principle ' 
comes out in our attitude to compensation. A man will not 
lightly give up a moral principle; nor will he lightly give up 
anything else that he regards as valuable. But our attitude 
towards giving up a moral principle differs from all other cases. 
If a man has a picture that he values very highly he may reject 
a low price and be more inclined to part with it if the bid is 
raised. But if a man refuses a bribe of ten pounds and you 
offer him a hundred, he might say: "You don't understand; 
it is not a question of how much; doing that sort of thing 
is against my moral principles". Indeed he must say this, 
if it is really a matter of moral principle, unless he can manage 
to bring the acceptance of the offer under some other moral 
principle. It is for this reason that Napoleon's dictum that 
every man has his price sounds so cynical ; it implies that no , 
man has any moral principles. \ 

(b) But consistency in action is not the only test of a man's ' 
moral principles. Although a man cannot claim that it is 
against his moral principles to be cowardly or mean if he 
regularly does cowardly or mean things, he can do such things 
occasionally and still justify this claim. His claim is justified 
if he is prepared to condemn his own actions and if he feels 
remorse. His moral principles are not those on which he always 
acts, but those which he acknowledges or avows and those 
about which he feels remorseful when he breaks them. His 
moral principles are those on which, in his more reflective 
moments, he honestly says that he would like to act; they are 
the moral principles of the person he is striving to become. 

308 ' 


I shall return to this point in the last section of this 

(c) A principle is not usually called a moral one unless the 
person who adopts it is prepared to apply it universally. If a 
man says that he does something as a matter of principle, he 
cannot (logically) make exceptions unless another moral 
principle is involved. However narrov/ in scope it may be, a 
moral principle must be applied to all cases that are alike in 
all relevant respects. If there are two people of roughly similar 
character, tastes, and habits, it may well be that a man likes 
one of them better than the other. If asked why, he may be 
unable to give a reason; he just happens to like Jones, although 
tie concedes that Smith is just as virtuous, charming, and 
amusing. And, although there is an oddity about his taste that 
might interest a psychologist, there is nothing logically odd 
about it. But he is abusing language if he says that it is a 
matter of moral principle with him to pay his debts and he 
pays Jones, while refusing to pay Smith, without being able 
to give any reason for the discrepancy. 

The logical fact that a pro-attitude is not called a 'moral 
principle' unless a man is prepared to universalize it has led 
some philosophers to suppose that it can be proved that we 
Dught to be impartial. But this is to commit the fallacy of 
deducing a moral injunction from a feature of moral language. 
A man who has no principles that he is prepared to apply 
impartially has no moral principles ; but we cannot prove that 
he ought to have any moral principles by pointing out how the 
phrase ' moral principles ' is used. 

(d) The fact that a man's moral principles are those which he 
acknowledges in his more reflective moments throws some 
light on the connexion between moral principles and rules. 
A man's moral principles are those on which he thinks he 
ought to act and the word 'ought', like all deontological 
words, is only used in connexion with rules and therefore 
in connexion with relatively long-range principles and 
policies that we avow and adopt in our more reflective 



Moreover these deontological words contextually imply a 
background of general agreement; so that, in deliberating 
about what to do, we tend to use the language of ' ought ' only 
in connexion with principles of action that we know to be 
generally approved. Now, for reasons given in chapter 14, 
moral codes never contain injunctions to people to pursue 
their own pleasure; and most moral rules are concerned with 
the welfare of others. These pervasive features of moral codes 
infect the logic of deontological words. It is odd to describe 
a man as a ' conscientious egoist ' or to say that pleasure-seeking 
is his highest moral principle, because people do not in fact use 
the language of ' ought ' when they are being deliberately and 
consistently selfish. And the reason for this is that it is hard 
to dissociate this word from its moorings in the language 
of advice, exhortation, and command. Nevertheless, if a 
man regularly decides that he ought (in the verdict-giving 
sense of ' ought ') to do whatever brings him pleasure or profit, 
his dominant pro-attitude is towards his own pleasure or 
profit. Whether or not we choose to call selfishness a moral 
principle with him, depends on the criterion we are using for 
the phrase 'moral principle'. If he behaves selfishly without 
acknowledging his wickedness and without feeling remorse, 
we could say that selfishness was one of his moral principles; 
and we hesitate to say this partly because he almost certainly 
does not address himself in the language of 'ought' (in the 
self-hortatory sense) and partly because we are reluctant to 
believe that he really is what he makes himself out to be. 


Can a man choose to act against his own moral principles or choose 
to change them? Some moral principles are fundamental in the 
sense that we can give no reasons for adopting them; they do 
not follow from any higher principles. And it follows that a 
man cannot, at the moment of choosing, question the validity 



of the principle on which he chooses to act. For to do this 
would be to criticize the principle in the light of a higher 
principle; and in that case the principle in question is not a 
fundamental one. A man cannot condemn the principle on 
which he acts unless he has a con-attitude towards it; and in 
that case it is not a fundamental pro-attitude. 

Now this seems to entail that a man cannot choose to act 
against his own moral principles, that he cannot choose to do 
what he knows to be wrong. But this is not so. Self-criticism 
is possible because, in criticizing my own character or conduct, 
I apply, not the principles on which I act, but the principles 
that I acknowledge on those occasions when there is no ques- 
tion of their being manifested or not manifested. I can, for 
example, think that I ought to be less greedy, vindictive, or 
sanctimonious than I am, and this implies a con- attitude to- 
wards these particular traits in my character. But I cannot 
(logically) condemn any of these vices in myself while at the 
same time exercising them. For if I behave vindictively while 
at the same time condemning myself for doing so, I am a 
weak-willed but not a vindictive person. If, on the other hand, 
I deliberately choose to do something vindictive, then I am 
a vindictive person ; and I can still claim that to be vindictive 
is against my principles only in the sense that, in my more 
reflective moments, I am prepared to condemn what I 

The answer to the question whether a man can choose to 
change his moral principles is partly logical, partly empirical. 
In the case of principles that are not fundamental there is no 
logical difficulty, since we adopt these for reasons and both can 
and should abandon them if we find that the reasons are bad 
reasons, although it may be in practice difficult to do so. 
Traditionally a large part of moral philosophy has consisted 
in the attempt to show that many moral principles are 
subordinate in this way to one or a few very general prin- 
ciples, such as the Golden Rule or the Greatest Happiness 

But, although there is no logical difficulty in the notion of 



trying to change a subordinate principle, there must, at any 
given moment, be some principles that are, here and now, 
fundamental moral principles for me. If this were not so, we 
could not talk about choosing or trying to change a principle, 
since this implies having a pro-attitude towards making the 
change. And it is here that the logical difficulty arises. 

To try to change a principle implies having a pro-attitude 
towards making the change, and this implies that the principle 
is not a fundamental one. But it does not follow from this that 
there are any moral principles that are unchangeable. The fact 
that it makes no sense for me to ask whether I ought to act 
on a certain principle that is for me a fundamental one has 
often been cited as a proof that there are self-evident principles. 
For is not to say that it is senseless to question the principle 
to say that it is self-evident? But this argument confuses 
the practical impossibility of asking a certain question at 
a certain time with the logical impossibility of asking it at any 
time ; and it also confuses the role of the advocate with that of 
the judge. 

So long as a man is considering whether or not to act in a 
certain way, he addresses himself in the split-personality 
language of ' you ought '. But sooner or later he must make up 
his mind; he must decide. No doubt perpetual indecision is 
logically possible; but in many cases not to decide is to take 
a momentous decision, since the situation alters and the oppor- 
tunity for choosing has passed. Moreover the logic of practical 
language is adapted to the practice of ordinary men, not to 
that of mental paralytics. 

Sooner or later, then, he must proceed to a verdict "This 
is what I ought to do; this is the principle on which I shall 
act". And it is logically impossible for him to question this 
decision only in the sense that, if he questioned it, he would be 
returning to the standpoint of the advocate and it would not be 
a decision. It does not follow that at some future time he might 
not reconsider the decision and wonder whether he had been 
right. But to question the morality of a decision or principle is 
to criticize or appraise it in the light of a higher principle. 



Could this principle be questioned in its turn? Unless it 

were tautologous (in which case it could not serve as a moral 

principle at all, since it would be compatible with every course 

of action), it could be. Self-guaranteeing moral principles 

are impossible; and the demand for them rests on the failure 

to notice that 'there must always be some moral principle 

that I cannot now question' does not entail 'There must 

ibe some moral principle that I cannot ever question'. Every 

j sentence must (logically) end with a full stop; but there is 

jno point in any sentence at which a full stop must (logically) 

Ibe put. 

j A man can, therefore, question the morality of his own 
principles and try to change them; but he cannot do so while 
applying them or if he has no pro-attitude towards making the 
change. Whether or not he can change them if these logical 
conditions are satisfied is an empirical question, to which the 
jonly answer is: " Sometimes. He may not always succeed; but 
he can always try". And since no one, not even the man him- 
self, knows the limits of what he can do if he tries, it is a 
question to which no more precise answer can be given. There 
are moral principles which it is difficult to imagine any man 
jwanting to change, because it is difficult to imagine what it 
would be like to adopt the contrary principle or to have a 
jpro-attitude towards adopting it. But we must not confuse 
^he difficulty of imagining something with its logical im- 
1 What sort of principles a man adopts will, in the end, depend 

ti his vision of the Good Life, his conception of the sort of 
orld that he desires, so far as it rests with him, to create, 
jindeed his moral principles just are this conception. The 
jbonception can be altered; perhaps he meets someone whose 
character, conduct, or arguments reveal to him new virtues 
that he has never even contemplated ; or he may do something 
Uncharacteristic and against his principles without choosing 
ko do it and, in doing it, discover how good it is. Moral 
values, like other values, are sometimes discovered accidentally. 
But the one thing he cannot do is to try to alter his conception 



of the Good Life; for it is ultimately by reference to this 
conception that all his choices are made. And the fact that 
he cannot choose to alter this conception neither shields 
him from blame nor disqualifies him from admiration. 





I HAVE tried in the course of this book to elucidate the con- 
cepts used in moral discourse and their connexions with each 
other; and I have referred at several points to very ancient 
controversies, between objectivists and subjectivists, deon- 
tologists and teleologists, libertarians and determinists. These 
controversies are perennial and it would be absurd to claim to 
settle them. It would be still more absurd to claim to settle 
them in favour of one side or the other, since it would be most 
unlikely for any theory to survive so long if it were wholly 

Nor could the issues be settled wholly by logical argument ; 
for they are partly matters of individual psychology. We may 
ask what ' we ' mean by a certain word ; but we do not all mean 
the same thing and, if we did, it would be impossible to under- 
stand why it is that, in a philosophical dispute, which is con- 
cerned with the meanings of words that are the common 
property of everybody, the points made by the protagonists 
on each side seem to their opponents so absurd, tenuous, and 
far-fetched. There may be, and usually is, a large measure of 
agreement on matters of morality; but the disagreement that 
persists is not wholly on matters of logic. 

Consider, for example, the reactions of two different people 
who are told that something they have done is wrong or 
immoral and who accept the criticism as just. For one of them 
the immediate effect of this criticism may be to make him 
think about the past. He thinks in terms of having broken a 
rule and his dominant emotion is shame, a sense of guilt or 
sin, a sense of his incapacity to live up to his ideals. Moreover 
it is almost certain that he will tend to think of the ideals he 



has failed to live up to and the rules he has broken as being 
not primarily his ideals and rules at all, but as being ' objective ', 
as belonging to a special order of reality that he did not 
create but vi^hich is imposed on him. In this way it is easier 
for him to understand hoM? he came to fail; and because it i 
is easier for him to understand his failure in this way it is 
more comforting. 

Another man will think primarily of the future; his domi- 
nant reactions will not be shame but a desire to put right the 
wrong he has done, if this is possible, and a desire to do better 
in future. For him morality is more a matter of what he ought 
to do now and in future than a matter of what he has done in 
the past. Shame and guilt, since they are concerned with a past 
that is dead and gone, will seem to him, not to be of the very 
essence of self-condemnation, but important only as spurs to 
future effort. And since he thinks more in terms of decisions 
for the future than of remorse for the past, he will not feel the 
same need to represent moral rules and ideals as imposed on 
him from without and will not derive the same comfort from 
so doing. Indeed it will seem to him to be a slavish attitude. 
For him rules and moral laws are important enough as guides 
in cases of doubt and as correctives for tendencies in his own 
conduct of which he disapproves in his reflective, self-critical 
moments; but they will not be so important as the policies 
and principles which he, as a free rational agent, chooses to 
adopt. Is it likely that two such men will agree about the 
meaning of the word 'wrong', especially if they make the 
logical assumption that it must refer to just one thing or 
have just one primary use ? 

But although these psychological considerations deserve 
more recognition than they usually receive, they must not be 
overestimated. To a very large extent the theories which seem 
to some philosophers paradoxical and far-fetched really are 
paradoxical and far-fetched. They misrepresent not only the 
way in which others use moral language but the way in which 
the theorist himself uses it. No man, for example, ever really 
believed that * good ' meant ' what I desire ' or that altruism 



was disguised selfishness or that obligations were denizens of a 
special, non-natural world. These are absurd, one-sided theories 
of which the only value lies in the fact that they emphasize ele- 
ments in moral language that might otherwise be overlooked. 
And they are arrived at by pursuing, often with great ingenuity 
and logical rigour, tendencies which are endemic in our actual 
language to the neglect of other tendencies not less important. 

For, whether it be a virtue or a vice, it is a fact that ordinary 
language is untidy ; almost everything that we say would have 
to be qualified by the phrase ' for the most part ' if it were not 
so obvious that this is the case that we can afford to dispense 
with the phrase in practice. But if, as philosophers sometimes 
do, we forget the untidiness of the logic of ordinary language, 
we shall find ourselves deducing consequences from uses of 
language that are incompatible with each other. And then we 
shall feel bound to reject one or other of the implications and, 
whichever we reject, we shall have a one-sided theory to offer. 
Philosophy is, for this reason, full of paradoxes; for a philo- 
sopher cannot but stress those features of language which seem 
to him important, and these may not seem to be the important 
ones to another philosopher. But as long as we see the para- 
doxes for what they are and know why it is that we reach them 
they do little harm. Sometimes, however, philosophical 
theories rest on sheer logical confusions and, at the risk of 
some repetition, I shall end by summarizing the most im- 
portant of these. 

(i) The most important and pervasive is that of transferring 
to discussions of moral discourse the logical concepts that 
have been successfully used to elucidate the discourse of mathe- 
matics or science. This has led philosophers to misrepresent 
knowing how to lead one's life as knowledge of theoretical 
truths, either about human nature or about a special realm of 
* values '. This error, combined with the realization that truths 
of fact do not entail imperatives and that neither truths of fact 
nor imperatives entail decisions, has led to the doctrine that 
moral words must stand for special entities and to the postu- 
lation of a special faculty to account for our knowledge of 



moral truths. The crucial difference between practical and 
theoretical discourse has been misrepresented as a difference 
between sets of objects described instead of represented as 
a difference in the role performed by different types of 

(2) Neglect of the distinction between the meaning of a 
word and the contextual background of its use and of the fact 
that an expession may have different implications in different 
contexts has led to partial analyses being offered as ' the mean- 
ing of X' or 'the way in which the word "X" is used', this 
latter fallacy being only the old fallacy of iinum nomen unum 
nominatum in modern dress. The logical liaisons of a given 
word are many and various and, although a given expression 
always has some contextual implications, what it implies in 
one context may be very different from what it implies in 
another. Once this fact is appreciated it is easier to see why 
what is a truism to one philosopher can seem to be a plain 
falsehood to another. The first has treated one set of logical 
liaisons as valid in all contexts, the second another set. 

(3) Teleological writers have often been accused of ' deny- 
ing morality altogether' or failing to account for moral obli- 
gation 'in the strict sense'. In a way this accusation is an 
impertinence that they rightly ignore ; but in another way it is 
just, since they have attempted to define deontological words 
in terms of purpose, happiness, desire, pleasure, or good. And 
in so doing they have confused logical questions about the 
meaning of the words 'ought', 'right', 'duty', 'just', and 
' obligatory ' with practical questions about what rules we ought 
to adopt and psychological questions about what pro-attitudes 
men in fact have. This procedure is rightly denounced as in- 
volving a failure to understand the peculiar part which 
deontological concepts actually play in choosing, advising, 
commanding, and exhorting. 

But the deontologist, while rightly emphasizing the con- 
nexion between ' ought ' and rules, has been led to treat desires 
and purposes as ' merely empirical ' concepts, the proper con- 
cern of the psychologist, not of the moral philosopher. And 



this has led him into the wildly extravagant assertion that all 
traditional moral philosophy rests on a mistake. By distin- 
guishing between the standpoint of the legislator and that of 
the judge we can do justice to the deontologist's insight that 
the question whether a verdict is just and the question whether 
the judge ought to pronounce it are questions of fact which 
have nothing to do with purposes, and also to the teleologist's 
insight into the fact that rules are made to serve the purposes 
of men. 

(4) The oversimplification involved in treating all pro-atti- 
tudes, that is to say all logically complete answers to the 
question ' What shall I do ? ', as ' desires ' or ' inclinations ' has 
resulted, not only in over-simple psychology, but also in the 
fatal tendency to represent all voluntary action as selfish; and 
this in turn is so flagrant a paradox that it has led to the treat- 
ment of the sense of duty as a non-natural force which must be 
a force, since it opposes the desires, and yet cannot be just 
another natural force, since it pronounces a verdict. But to 
suppose that, the sense of duty apart, all men would act 
selfishly is to suppose something for which we can have no 
evidence, since we know of no societies in which there is no 
concept of obligation ; and it is something which is quite cer- 
tainly untrue of men as we know them. 


Moral philosophy is a practical science; its aim is to answer 
questions in the form 'What shall I do?'. But no general 
answer can be given to this type of question. The most a moral 
philosopher can do is to paint a picture of various types of life 
in the manner of Plato and ask which type of life you really 
want to lead. But this is a dangerous task to undertake. For the 
type of life you most want to lead will depend on the sort 
of man you are. Decisions and imperatives do not follow logi- 
cally from psychological or biological descriptions; but the 
sort of life that will in fact be satisfactory to a man will depend 



on the sort of man that he is. Generalization is possible only 
in so far as men are psychologically and biologically similar. 
There are some types of life that we can say outright that no 
man would find satisfactory; but practical advice is not neces- 
sary when it is obvious. In cases which are difficult to decide 
it is vain, presumptuous, and dangerous to try to answer these 
questions without a knowledge both of psychology and of the 
individual case. 

My purpose has been the less ambitious one of showing how 
the concepts that we use in practical discourse, in deciding, 
choosing, advising, appraising, praising and blaming, and 
selecting and rejecting moral rules are related to each other. 
The questions ' What shall I do ? ' and ' What moral principles 
should I adopt?' must be answered by each man for himself; 
that at least is part of the connotation of the word ' moral '. 



The principal references are indicated by heavy type. 
h 287, 

Addiction, 265-7, 2: 

Advising, 145-59, 162, 188-9, 

192, 198, 225, 254 
Altruism, 134-5 > 140-4. 246, 

253, 255 
Analytic/Synthetic, 32, 75-9» 

98-9, 102, 107, 270 
Appraising, 12, 164, 170-7, 

180-1, 288, 290, 301 
Approval, 39, 60, 95-6, 176- 

81, 304 
Aristotle, 13, 15, 19, 37, 126, 

235 'n, 259, 292 n 
A-words (-sentences), 70-5, 

84-91, 118-20, 146-51, 

154-5, I 60-1 

Beneficence, 231, 249 
Benevolence, 214, 231, 246, 

Blame (-worthy), 264, 270-1, 

285-9, 291-306, 314 
Broad, C. D., 24-7, 33, Si, 56 
Butler, Bishop, 13, 218, 223, 

229, 244 

Campbell, C. A., 279-81, 

Can/Could, 201-3, 210, 273- 

81, 283-6, 291, 294-300, 


Character, 123-4, 270, 280-4, 

287-90, 297-306 
Choosing, 100-4, 1 16-19, 

160-2, 168-9, 200-5, 223, 

262-5, 278-82, 287-90, 

Clarke, S., 47-8, 50 
Codes, moral and legal, 18-20, 

188-9, 209-11, 214, 227, 

252, 256-7, 272, 288, 310 
Conscience, 151, 197, 218, 

221, 242-4, 260-9 
Conscientiousness, 213-14, 

245-59, 264, 289-90 
Contextual Implication, 80-7, 

96-9, I 17-21, 146-55, 

160-9, 186, 192-9, 223, 

276, 318 
Craving, 107, 265-6 

Deciding, 101-2, 152, 260-3, 

312, 317-19 
Denoting, 31-3, 62-9 
Deonotological/Teleological , 

13, 133, 181-2, 188, 216- 

25, 245, 318-19 
Desire, 106-11, 134-43, 151, 

205-7, 261-4, 279-80, 

287-8, 296-7, 319 
Disagreement, moral, 44-7, 

55-6, 193-7, 253 
Duty, 13, 34-5, 198-215, 




Duty and Inclination, 134,200, Identity, personal, 283 
211-12, 246, 252, 262-5, Impartiality, 309 
279-80 (see also Sense of Incest, 235 

Duty) Inductive arguments, 88, 

Egoism, 14 n, 119", 133-S, Introspection, 40, 129-31, 281 

141-4, 199, 221-2, 234 Intuition, 23, 44-7, 55, 76-8 I 

Enjoying, 127-32, 137-41 Intuitionism, 24, 30, 32, 36- j 

Epicurus, 13, 138 47. 48-50, 56-60, 79, 133,' 

217-19, 237 I 

Fittingness, 120-1, 186-7, 

285, 302 Janus-words, 100, 107, 109, 
Frankfort, H., 14 n 120, 167, 223, 234, 237, 

Free, 205-9, 263-4, 267, 279- 241, 251 

83 Joseph, H. W. B., 63 n 

Freud, S., 106, 135, 235/2 Judges, 236-9, 261-2, 272-3, 

312, 319 

Gay, J., 220-2 J"^*^'^^' ^31-2, 236-9, 270-3, 
God, 33, 37, 192-3, 218, 282 ^99 

Good (-ness), 32-4, 63-5, 

75-9, 95-6, 160-82, 233, Kant, I., 13, 133-4, 245-8 


Good Life, the, 12-15,61,220, Law, laws, 227-8, 236-9, 

3^3-14 270-3, 291-3, 299 (see also 

G-words (-sentences), 70, 73, Codes) 

91, 120, 151-5, 190-1, Legislator, 236-9, 272-3, 319 

222-5, 270-1 Libertarianism, 278-90, 303, 


Hart, H. L. A., 292 n Locke, J., 76 

Hedonism, 32, 133-40 Logic, 15, 20-2, 63, 317 

Hobbes, T., 13, 17, 106, Logic of adjectives, 61-74 

iio-ii, 115, 134, 135 M, Logic of modal verbs, 273-8, 

220, 228, 238 281 

Hume, D., 36-7, 40, 100, 117, Logic of sentences, 75-91 

176, 220, 229, 244 Logical limits, 81, 115, 132, 
Hutcheson, F., 159 «, 176 188, 220, 238, 245, 249 



Logically odd, 72-5, 80, 83-5, 
I03-7, III, 113, 154, 
161-3, 169, 177-8, 251, 
254, 278 

Logic and Psychology, 105-6, 
114-15, 135^. 181-2, 219, 

Luke, St, 14 n 

Mandeville, B. de, 159 n 
Masochism, 109, 132 
Mathematics, 17, 34, 47, 97 n 
Mill, J. S., 137-8, 282 
Moore, G. E., 32-4, 36, 64-5, 

73-5, 171, 222 
Mortimer, Bishop, 37 n 
Motives, 106, III, 122-32, 

204-8, 226-7, 245, 251-7, 


Natural Law, 237, 272-3 
Naturalistic Fallacy, the, 32-3, 

36, 71, 75, 176, 180-2 
Non-natural, 33, 39-43, 48, 

61,65,95, 216,237, 319 

Objective/Subjective, 26, 43-6, 
49-60, 96, 174, 178-80, 316 

Obligation , 2 o o- 1 2 , 2 1 9 , 240-4 
(see also Duty, Ought) 

Ought, 37-8, 41-3, 182, 185- 
92, 215-19, 223-5, 229, 
252-4, 267-9, 284, 310 

Ought, I and You ought, 
i93-7» 198-9, 240-4, 260-2, 

Oxford English Dictionary, 
95, 163 

Pain, 107-11, 130-2, 214, 271, 

299, 304 
Paton, H. J., 134, 246-7, 

248 n 
Plato, 13, 16, 23, 133, 198-9, 

218, 223, 228, 242, 319 
Pleasure, 17, 106, 131, 135-7, 

138-43, 211, 214, 304 
Practical/Theoretical, 11-22, 

23-5, 36, 41, 61-3, 69, 95- 

104, 185-93, 224-5, 317-20 
Praising, 98, 100, 163, 168-9, 

248-51, 256-7, 270, 300-5 
Prichard, H. A., 34-5, 38-9, 

Principles, moral, 178, 197, 

233, 254, 259, 261-9, 284, 

288-9, 293-4, 298, 306-14, 

Pro -attitudes, 11 2-21, 146-9, 

155-8, 178-82, 186-92, 

249-51, 255-7, 271, 301-13, 

Promising, 211, 232, 240-2, 

Pro-words, 1 13-14, 128-32, 

Psychology, 15-16, 25-7, 182, 

262, 281, 315-20 (see also 


Raglan, Lord, 235 n 
Raphael, D. D., 176, 244 w 
Regulus, 221-2 



Remorse, 25-6, 39, 308-11 
Responsibility, 263, 270, 278, 

283, 299-300, 303 
Rochefoucauld, Due de la, 135 
Romantic Movement, the, 289 
Ross, Sir W. David, 27-32, 

39, 50. 243, 252-4 
Rules, 14, 17-19, 121, 186-9, 

216-19, 226-44, 270-2, 316 

(see also Codes) 
Ryle, G., 125-35 

Saints, 138, 259 

Secondary uses, 81, 158, 166, 

173, 181, 189, 191, 194 
Self, the, 263, 282-3, 288 
Self-criticism, 25, 265, 308-11 
Self-determination, 282-3 
Selfish (-ness), 134-5, i42-4> 
158, 198-g, 214, 248, 258, 
310, 319 (see also Egoism) 
Sense of Duty, the, 1 15-16, 
133-5, 140, 143, 227, 244-6, 
258-9, 287,307, 319 
Socrates, 218, 264 
Spinoza, B., 13, 17, 138, 264 
Standards, 170-5, 179-81 
Subjective (see Objective) 
Synthetic (see Analytic) 

Teleological (see Deontolo- 

Temptation, 266-7, 282, 284, 

Theoretical (see Practical) 
Thrasymachus, 198, 218, 244 
Thrift, 187, 250 1 

Toulmin, S. E., 52 n, 89 n 
Training, moral, 243, 303-4 
Trying, 265, 280, 285-8, 


Utilitarian (-ism), 17, 232-9, 

Verdicts, 164, 170, 177, 185, 
190, 236-8, 261, 268, 283, 
292, 303-5, 319 

Virtues, names of, 100, 163, 
168, 301-2, 307 

Virtues, natural and artifi- 
cial, 213, 248-52, 255-9 

Vivas, E., 136-7 

Voting, 98-9, loi 

Weakness, moral, 265-9, 284, 

Westermark, E., 27 
Wickedness, 265-6, 284, 298, 

306, 310 

Some other books on philosophy published by 

Penguins are described on the 

follozving pages 



Plato (427-347 B.C.), finally disillusioned by contemporary 
politics after the execution of Socrates, showed in his writ- 
ings the enormous influence of that great philosopher. The 
Republic, his treatise on an ideal state, was the first of its 
kind in European thought. For Plato, political science was 
the science of the soul, and included moral science. The 
Republic's emphasis on the right education for rulers, the 
prevalence of justice, and harmony between all classes of 
society, is as strong as its condemnation of democracy, 
which Plato considered encouraged bad leadership. 

also available 

The Last Days of Socrates 

Protagoras and Meno 


A. J. Ayer 

What is knowledge and how do we knozo things ? Moreover, 
how do we know that we know them, in view of the doubts 
which the philosophical sceptic casts on our grasp of facts ? 
The presentation of the sceptic's arguments leads here to a 
general discussion of the topic of scepticism and certainty. 
This is followed by a detailed examination of the philoso- 
phical problems of perception, memory, and our knowledge 
of other minds, which occupies the greater part of the book. 
In the course of the discussion. Professor Ayer has also 
attempted to throw light upon the nature of philosophical 
method and upon some of the problems connected with 
time, causality, and personal identity. 

'The book is so thorough, so penetrating, vigorous and 
up to date, and at the same time so readable and manage- 
able, that it is unlikely to have a serious competitor for 
many years' - Spectator 

Kenneth Walker 

This is an unusually ambitious and wide-ranging book in 
our specialist age. But man is a complex creature who defies 
any simple, one-sided analysis. Kenneth Walker's diagnosis 
is that of a consulting surgeon with a broad culture and 
a more than medical concern for his fellow men. He 
shows how the biological approach can reveal much about 
the workings of the body-mind. But a profound mystery 
remains, and the author calls in philosophy and religion to 
illuminate that part of man which a solely mechanistic 
biology leaves in darkness. 

This Pelican edition contains an entirely new chapter on 
existentialism, and has been thoroughly revised throughout. 
'A striking and stimulating book. It should be read by all 
who are inclined to agree that the present crisis in Western 
civilization is a symptom not a cause, and even more by 
those to whom that idea has not yet occurred' - Sunday 

also available 

Human Physiology 

The Physiology of Sex 

Sex and Society 
Patients and Doctors 

Patrick Gardiner 

The revolutionary nature of some of Schopenhauer's ideas 
anticipating the insights of Freud, has only become fully 
apparent long after his death in i860. This book has been 
written by an Oxford philosopher in the belief that the 
current neglect of Schopenhauer's philosophy, which exer- 
cised a far-reaching influence on the thought of his age, is 

An admirer of Kant's Critique of Pure Reason, Schopen- 
hauer was very conscious of the snags in some traditional 
conceptions of metaphysics, and his own critical acumen 
set him apart from many of his German contemporaries. 
The original and imaginative system he created, and parti- 
cularly his theory of the will, though highly speculative, 
offered a strong challenge to many old assumptions which 
had seldom been questioned before, 

Patrick Gardiner's new study outlines and evaluates the 
main features of Schopenhauer's thought, explains their 
historical importance, and at the same time relates them to 
problems which occupy philosophers today. 


Stuart Hai7ip shire 

Spinoza is probably the most difficult of all the great philoso- 
phers to understand at first reading and without some 
general introduction to his methods and purposes. He is also 
supremely worth understanding, as one of the greatest of 
the metaphysicians who have tried to construct a coherent 
account of the universe and of man's place within it. His 
system embraces the whole range of the traditional prob- 
lems of philosophy including the problems of a creator and 
his creation, the relation of mind and body, the freedom of 
the will, the nature and limits of human knowledge, and also 
the problems of moral and political philosophy ; all the main 
themes of philosophical speculation can be found in his 
work. This book is designed as a general introduction which 
explains the connexions between the various aspects of 
Spinoza's philosophy, between his logic and his metaphysics 
and again between his metaphysics and his moral and poli- 
tical doctrines. 

'This book is a model of its kind . . . Mr Hampshire's criti- 
cal summary of the system is admirable : it is full, clear - or 
as clear as the subject allows - agreeably written, wastes no 
words and shirks no difficulties' - The Times Literary 

Richard Wollheim 

Francis Herbert Bradley (i 846-1 924) was the greatest 
British metaphysician of the nineteenth century. Generally 
characterized as an Idealist, he certainly drew heavily upon 
German metaphysics since Kant, but no simple classification 
or historical label does justice to his rich and imaginative 
scheme of thought. 

In his own day Bradley effected a revolution in British 
philosophy, and enjoyed a fame that is given to few philo- 
sophers in their lifetime. Since then, however, he has fallen 
an obvious victim to the anti-metaphysical tendencies of 
twentieth-century philosophy. The aim of the present study 
is to redress the balance. The author argues that a great deal 
of modern criticism of Bradley is based on misconception, 
and he shows how at many points Bradley was close to the 
interests and preoccupations of the present day. Bradley's 
vast metaphysical system does not rest, as is commonly 
supposed, on extravagant speculation, but on a theory of 
thought and symbolism; and this theory was an attempt to 
deal with certain very general problems about meaning with 
which contemporary philosophers and logicians find them- 
selves permanently engaged. If in large part the answers 
that he gave to these problems are very different from those 
which nowadays are given to them, this is an additional 
reason for not neglecting his contribution. 

Though one of the great masters of English prose, 
Bradley is not an easy philosopher to follow, and the present 
study is the first systematic attempt to give the outline and 
much of the detail of his thought in a way that will be 
intelligible both to the student of philosophy and to the 
general reader. 


S. Korner 

Immanuel Kant, who was born in 1724 and died in 1804, 
is by universal consent one of the greatest philosophical 
thinkers of the modern Western world. He combined the 
rare gifts of analytical acumen and constructive imagination 
with the still rarer gift of keeping the balance between the 
two. There are hardly any of the perennial problems of 
philosophy about which he did not think deeply and closely 
in terms of experience, and which have not been illuminated 
by being placed in the light of his philosophical system as a 
whole. Perhaps no thinker ever influenced his successors 
more. Even the writings of those among them who opposed 
or oppose him, or who have never properly studied his work, 
abound in thoughts which he was the first to formulate. 

This introduction offers an outline of Kant's system, 
one of its chief aims being to show that his problems and 
solutions are not merely of historical interest, but that they 
concern everybody who makes statements of fact and judge- 
ments of value. 



Gordon Lejf 

In the history of ideas, there is a tendency for the old attitude 
to persist that between the Roman Empire and the Renais- 
sance there was nothing of importance. It is now generally 
accepted that, in fact, the Middle Ages had their own 
distinctive civilization, and even their doctrines were more 
than the mere embellishments of Catholic dogma. This 
book, in tracing their development over the thousand years 
from St Augustine to Ockham, shows that the picture was 
far from static or uniform. The author treats the subject 
historically. He begins with a discussion of the special nature 
of the medieval outlook and the different sources from which 
its thinkers drew. The book is divided into three parts, 
corresponding to the main phases in medieval life as well 
as thought, each preceded by a short historical introduction 
for those unacquainted with the period. Close connexion is 
observed between the thinkers and their cultural milieu ; and 
educational developments, such as the rise of the univer- 
sities, are considered throughout. 

' The best short account of medieval philosophy available ' - 
Classical Journal 

'An admirable presentation of the bewildering variety of 
medieval teaching ' - History 


Bertrand Russell 

*It all depends what you mean by . . .' The late Professor 
C. E. M. Joad sometimes amused and sometimes irritated 
listeners to the Brains Trust with this invariable comment on 
the questions asked. But he was merely in accord with the 
modern trend of philosophy, which nowadays stresses the 
importance of close linguistic analysis. 

In this fundamental inquiry Bertrand Russell, author of A 
History of Western Philosophy and probably the best-known 
philosopher of our time, is concerned above all with langu- 
age, as he examines the processes of thought, expression, and 
knowledge with mathematical precision. What is the relation 
between a belief and a sentence in which that belief is ex- 
pressed ? How are we to set about defining 'knowledge' and 
'truth' and relating these concepts to our own experience ? 

In answering such questions Bertrand Russell applies the 
methods of psychology as well as logic. Whilst his answers 
are often complex - as, for precision, they must be - his argu- 
ments are at all times both lucid and elegant. 

Also available 

Has Man a Future? 

Unarmed Victory 

Nightmares of Eminent Persons 

Satan in the Suburbs 


Henri Frankfort and others 

Before philosophy, in the strict sense of abstract, critical, 
and methodical thought, came into being, man's specula- 
tions, when they turned to the perennial problems of self 
and the universe, were expressed in myths. These myths 
are not mere stories, nor do they merely disguise an abstract 
truth. They represent a peculiar form of concrete thought, 
which should be analysed before the attempt is made to 
understand the mind of Ancient Man, his moral and relig- 
ious preoccupations, ^ 

The authors, who have concentrated on the two oldest 
civilizations known - those of Egypt and Mesopotamia - 
have not shirked the first task. Just because they aimed at 
a deeper understanding of the Egyptian and Mesopotamian 
points of view in matters of life and death, the function of 
the State, and the natmre of the phenomenal world, they 
have taken the problem of myth seriously. 

They have discussed the emergence of Greek philosophy 
in order to stress the gulf which separates the habits of 
thought of the Greeks from those of their predecessors. 

For a complete list of books available please write to Penguin Boohs 
whose address can be found on the back of the title page 


Mrrt 1 c c^u; 

M 2 ^ ?M3 

A fine of 25 cents per day is charged when this item is overdue. | 



3 1262 05383 7463 N^'4-^t 

c 'X